This commit is contained in:
Waylon Walker 2022-03-31 20:20:07 -05:00
commit 38355d2442
No known key found for this signature in database
GPG key ID: 66E2BF2B4190EFE4
9083 changed files with 1225834 additions and 0 deletions

File diff suppressed because it is too large Load diff

View file

@ -0,0 +1,3 @@
from black import patched_main
patched_main()

View file

@ -0,0 +1,334 @@
"""Builds on top of nodes.py to track brackets."""
from dataclasses import dataclass, field
import sys
from typing import Dict, Iterable, List, Optional, Tuple, Union
if sys.version_info < (3, 8):
from typing_extensions import Final
else:
from typing import Final
from blib2to3.pytree import Leaf, Node
from blib2to3.pgen2 import token
from black.nodes import syms, is_vararg, VARARGS_PARENTS, UNPACKING_PARENTS
from black.nodes import BRACKET, OPENING_BRACKETS, CLOSING_BRACKETS
from black.nodes import MATH_OPERATORS, COMPARATORS, LOGIC_OPERATORS
# types
LN = Union[Leaf, Node]
Depth = int
LeafID = int
NodeType = int
Priority = int
COMPREHENSION_PRIORITY: Final = 20
COMMA_PRIORITY: Final = 18
TERNARY_PRIORITY: Final = 16
LOGIC_PRIORITY: Final = 14
STRING_PRIORITY: Final = 12
COMPARATOR_PRIORITY: Final = 10
MATH_PRIORITIES: Final = {
token.VBAR: 9,
token.CIRCUMFLEX: 8,
token.AMPER: 7,
token.LEFTSHIFT: 6,
token.RIGHTSHIFT: 6,
token.PLUS: 5,
token.MINUS: 5,
token.STAR: 4,
token.SLASH: 4,
token.DOUBLESLASH: 4,
token.PERCENT: 4,
token.AT: 4,
token.TILDE: 3,
token.DOUBLESTAR: 2,
}
DOT_PRIORITY: Final = 1
class BracketMatchError(Exception):
"""Raised when an opening bracket is unable to be matched to a closing bracket."""
@dataclass
class BracketTracker:
"""Keeps track of brackets on a line."""
depth: int = 0
bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = field(default_factory=dict)
delimiters: Dict[LeafID, Priority] = field(default_factory=dict)
previous: Optional[Leaf] = None
_for_loop_depths: List[int] = field(default_factory=list)
_lambda_argument_depths: List[int] = field(default_factory=list)
invisible: List[Leaf] = field(default_factory=list)
def mark(self, leaf: Leaf) -> None:
"""Mark `leaf` with bracket-related metadata. Keep track of delimiters.
All leaves receive an int `bracket_depth` field that stores how deep
within brackets a given leaf is. 0 means there are no enclosing brackets
that started on this line.
If a leaf is itself a closing bracket, it receives an `opening_bracket`
field that it forms a pair with. This is a one-directional link to
avoid reference cycles.
If a leaf is a delimiter (a token on which Black can split the line if
needed) and it's on depth 0, its `id()` is stored in the tracker's
`delimiters` field.
"""
if leaf.type == token.COMMENT:
return
self.maybe_decrement_after_for_loop_variable(leaf)
self.maybe_decrement_after_lambda_arguments(leaf)
if leaf.type in CLOSING_BRACKETS:
self.depth -= 1
try:
opening_bracket = self.bracket_match.pop((self.depth, leaf.type))
except KeyError as e:
raise BracketMatchError(
"Unable to match a closing bracket to the following opening"
f" bracket: {leaf}"
) from e
leaf.opening_bracket = opening_bracket
if not leaf.value:
self.invisible.append(leaf)
leaf.bracket_depth = self.depth
if self.depth == 0:
delim = is_split_before_delimiter(leaf, self.previous)
if delim and self.previous is not None:
self.delimiters[id(self.previous)] = delim
else:
delim = is_split_after_delimiter(leaf, self.previous)
if delim:
self.delimiters[id(leaf)] = delim
if leaf.type in OPENING_BRACKETS:
self.bracket_match[self.depth, BRACKET[leaf.type]] = leaf
self.depth += 1
if not leaf.value:
self.invisible.append(leaf)
self.previous = leaf
self.maybe_increment_lambda_arguments(leaf)
self.maybe_increment_for_loop_variable(leaf)
def any_open_brackets(self) -> bool:
"""Return True if there is an yet unmatched open bracket on the line."""
return bool(self.bracket_match)
def max_delimiter_priority(self, exclude: Iterable[LeafID] = ()) -> Priority:
"""Return the highest priority of a delimiter found on the line.
Values are consistent with what `is_split_*_delimiter()` return.
Raises ValueError on no delimiters.
"""
return max(v for k, v in self.delimiters.items() if k not in exclude)
def delimiter_count_with_priority(self, priority: Priority = 0) -> int:
"""Return the number of delimiters with the given `priority`.
If no `priority` is passed, defaults to max priority on the line.
"""
if not self.delimiters:
return 0
priority = priority or self.max_delimiter_priority()
return sum(1 for p in self.delimiters.values() if p == priority)
def maybe_increment_for_loop_variable(self, leaf: Leaf) -> bool:
"""In a for loop, or comprehension, the variables are often unpacks.
To avoid splitting on the comma in this situation, increase the depth of
tokens between `for` and `in`.
"""
if leaf.type == token.NAME and leaf.value == "for":
self.depth += 1
self._for_loop_depths.append(self.depth)
return True
return False
def maybe_decrement_after_for_loop_variable(self, leaf: Leaf) -> bool:
"""See `maybe_increment_for_loop_variable` above for explanation."""
if (
self._for_loop_depths
and self._for_loop_depths[-1] == self.depth
and leaf.type == token.NAME
and leaf.value == "in"
):
self.depth -= 1
self._for_loop_depths.pop()
return True
return False
def maybe_increment_lambda_arguments(self, leaf: Leaf) -> bool:
"""In a lambda expression, there might be more than one argument.
To avoid splitting on the comma in this situation, increase the depth of
tokens between `lambda` and `:`.
"""
if leaf.type == token.NAME and leaf.value == "lambda":
self.depth += 1
self._lambda_argument_depths.append(self.depth)
return True
return False
def maybe_decrement_after_lambda_arguments(self, leaf: Leaf) -> bool:
"""See `maybe_increment_lambda_arguments` above for explanation."""
if (
self._lambda_argument_depths
and self._lambda_argument_depths[-1] == self.depth
and leaf.type == token.COLON
):
self.depth -= 1
self._lambda_argument_depths.pop()
return True
return False
def get_open_lsqb(self) -> Optional[Leaf]:
"""Return the most recent opening square bracket (if any)."""
return self.bracket_match.get((self.depth - 1, token.RSQB))
def is_split_after_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> Priority:
"""Return the priority of the `leaf` delimiter, given a line break after it.
The delimiter priorities returned here are from those delimiters that would
cause a line break after themselves.
Higher numbers are higher priority.
"""
if leaf.type == token.COMMA:
return COMMA_PRIORITY
return 0
def is_split_before_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> Priority:
"""Return the priority of the `leaf` delimiter, given a line break before it.
The delimiter priorities returned here are from those delimiters that would
cause a line break before themselves.
Higher numbers are higher priority.
"""
if is_vararg(leaf, within=VARARGS_PARENTS | UNPACKING_PARENTS):
# * and ** might also be MATH_OPERATORS but in this case they are not.
# Don't treat them as a delimiter.
return 0
if (
leaf.type == token.DOT
and leaf.parent
and leaf.parent.type not in {syms.import_from, syms.dotted_name}
and (previous is None or previous.type in CLOSING_BRACKETS)
):
return DOT_PRIORITY
if (
leaf.type in MATH_OPERATORS
and leaf.parent
and leaf.parent.type not in {syms.factor, syms.star_expr}
):
return MATH_PRIORITIES[leaf.type]
if leaf.type in COMPARATORS:
return COMPARATOR_PRIORITY
if (
leaf.type == token.STRING
and previous is not None
and previous.type == token.STRING
):
return STRING_PRIORITY
if leaf.type not in {token.NAME, token.ASYNC}:
return 0
if (
leaf.value == "for"
and leaf.parent
and leaf.parent.type in {syms.comp_for, syms.old_comp_for}
or leaf.type == token.ASYNC
):
if (
not isinstance(leaf.prev_sibling, Leaf)
or leaf.prev_sibling.value != "async"
):
return COMPREHENSION_PRIORITY
if (
leaf.value == "if"
and leaf.parent
and leaf.parent.type in {syms.comp_if, syms.old_comp_if}
):
return COMPREHENSION_PRIORITY
if leaf.value in {"if", "else"} and leaf.parent and leaf.parent.type == syms.test:
return TERNARY_PRIORITY
if leaf.value == "is":
return COMPARATOR_PRIORITY
if (
leaf.value == "in"
and leaf.parent
and leaf.parent.type in {syms.comp_op, syms.comparison}
and not (
previous is not None
and previous.type == token.NAME
and previous.value == "not"
)
):
return COMPARATOR_PRIORITY
if (
leaf.value == "not"
and leaf.parent
and leaf.parent.type == syms.comp_op
and not (
previous is not None
and previous.type == token.NAME
and previous.value == "is"
)
):
return COMPARATOR_PRIORITY
if leaf.value in LOGIC_OPERATORS and leaf.parent:
return LOGIC_PRIORITY
return 0
def max_delimiter_priority_in_atom(node: LN) -> Priority:
"""Return maximum delimiter priority inside `node`.
This is specific to atoms with contents contained in a pair of parentheses.
If `node` isn't an atom or there are no enclosing parentheses, returns 0.
"""
if node.type != syms.atom:
return 0
first = node.children[0]
last = node.children[-1]
if not (first.type == token.LPAR and last.type == token.RPAR):
return 0
bt = BracketTracker()
for c in node.children[1:-1]:
if isinstance(c, Leaf):
bt.mark(c)
else:
for leaf in c.leaves():
bt.mark(leaf)
try:
return bt.max_delimiter_priority()
except ValueError:
return 0

View file

@ -0,0 +1,99 @@
"""Caching of formatted files with feature-based invalidation."""
import os
import pickle
from pathlib import Path
import tempfile
from typing import Dict, Iterable, Set, Tuple
from platformdirs import user_cache_dir
from black.mode import Mode
from _black_version import version as __version__
# types
Timestamp = float
FileSize = int
CacheInfo = Tuple[Timestamp, FileSize]
Cache = Dict[str, CacheInfo]
def get_cache_dir() -> Path:
"""Get the cache directory used by black.
Users can customize this directory on all systems using `BLACK_CACHE_DIR`
environment variable. By default, the cache directory is the user cache directory
under the black application.
This result is immediately set to a constant `black.cache.CACHE_DIR` as to avoid
repeated calls.
"""
# NOTE: Function mostly exists as a clean way to test getting the cache directory.
default_cache_dir = user_cache_dir("black", version=__version__)
cache_dir = Path(os.environ.get("BLACK_CACHE_DIR", default_cache_dir))
return cache_dir
CACHE_DIR = get_cache_dir()
def read_cache(mode: Mode) -> Cache:
"""Read the cache if it exists and is well formed.
If it is not well formed, the call to write_cache later should resolve the issue.
"""
cache_file = get_cache_file(mode)
if not cache_file.exists():
return {}
with cache_file.open("rb") as fobj:
try:
cache: Cache = pickle.load(fobj)
except (pickle.UnpicklingError, ValueError, IndexError):
return {}
return cache
def get_cache_file(mode: Mode) -> Path:
return CACHE_DIR / f"cache.{mode.get_cache_key()}.pickle"
def get_cache_info(path: Path) -> CacheInfo:
"""Return the information used to check if a file is already formatted or not."""
stat = path.stat()
return stat.st_mtime, stat.st_size
def filter_cached(cache: Cache, sources: Iterable[Path]) -> Tuple[Set[Path], Set[Path]]:
"""Split an iterable of paths in `sources` into two sets.
The first contains paths of files that modified on disk or are not in the
cache. The other contains paths to non-modified files.
"""
todo, done = set(), set()
for src in sources:
res_src = src.resolve()
if cache.get(str(res_src)) != get_cache_info(res_src):
todo.add(src)
else:
done.add(src)
return todo, done
def write_cache(cache: Cache, sources: Iterable[Path], mode: Mode) -> None:
"""Update the cache file."""
cache_file = get_cache_file(mode)
try:
CACHE_DIR.mkdir(parents=True, exist_ok=True)
new_cache = {
**cache,
**{str(src.resolve()): get_cache_info(src) for src in sources},
}
with tempfile.NamedTemporaryFile(dir=str(cache_file.parent), delete=False) as f:
pickle.dump(new_cache, f, protocol=4)
os.replace(f.name, cache_file)
except OSError:
pass

View file

@ -0,0 +1,285 @@
import sys
from dataclasses import dataclass
from functools import lru_cache
import re
from typing import Iterator, List, Optional, Union
if sys.version_info >= (3, 8):
from typing import Final
else:
from typing_extensions import Final
from blib2to3.pytree import Node, Leaf
from blib2to3.pgen2 import token
from black.nodes import first_leaf_column, preceding_leaf, container_of
from black.nodes import STANDALONE_COMMENT, WHITESPACE
# types
LN = Union[Leaf, Node]
FMT_OFF: Final = {"# fmt: off", "# fmt:off", "# yapf: disable"}
FMT_SKIP: Final = {"# fmt: skip", "# fmt:skip"}
FMT_PASS: Final = {*FMT_OFF, *FMT_SKIP}
FMT_ON: Final = {"# fmt: on", "# fmt:on", "# yapf: enable"}
COMMENT_EXCEPTIONS = {True: " !:#'", False: " !:#'%"}
@dataclass
class ProtoComment:
"""Describes a piece of syntax that is a comment.
It's not a :class:`blib2to3.pytree.Leaf` so that:
* it can be cached (`Leaf` objects should not be reused more than once as
they store their lineno, column, prefix, and parent information);
* `newlines` and `consumed` fields are kept separate from the `value`. This
simplifies handling of special marker comments like ``# fmt: off/on``.
"""
type: int # token.COMMENT or STANDALONE_COMMENT
value: str # content of the comment
newlines: int # how many newlines before the comment
consumed: int # how many characters of the original leaf's prefix did we consume
def generate_comments(leaf: LN, *, preview: bool) -> Iterator[Leaf]:
"""Clean the prefix of the `leaf` and generate comments from it, if any.
Comments in lib2to3 are shoved into the whitespace prefix. This happens
in `pgen2/driver.py:Driver.parse_tokens()`. This was a brilliant implementation
move because it does away with modifying the grammar to include all the
possible places in which comments can be placed.
The sad consequence for us though is that comments don't "belong" anywhere.
This is why this function generates simple parentless Leaf objects for
comments. We simply don't know what the correct parent should be.
No matter though, we can live without this. We really only need to
differentiate between inline and standalone comments. The latter don't
share the line with any code.
Inline comments are emitted as regular token.COMMENT leaves. Standalone
are emitted with a fake STANDALONE_COMMENT token identifier.
"""
for pc in list_comments(
leaf.prefix, is_endmarker=leaf.type == token.ENDMARKER, preview=preview
):
yield Leaf(pc.type, pc.value, prefix="\n" * pc.newlines)
@lru_cache(maxsize=4096)
def list_comments(
prefix: str, *, is_endmarker: bool, preview: bool
) -> List[ProtoComment]:
"""Return a list of :class:`ProtoComment` objects parsed from the given `prefix`."""
result: List[ProtoComment] = []
if not prefix or "#" not in prefix:
return result
consumed = 0
nlines = 0
ignored_lines = 0
for index, line in enumerate(re.split("\r?\n", prefix)):
consumed += len(line) + 1 # adding the length of the split '\n'
line = line.lstrip()
if not line:
nlines += 1
if not line.startswith("#"):
# Escaped newlines outside of a comment are not really newlines at
# all. We treat a single-line comment following an escaped newline
# as a simple trailing comment.
if line.endswith("\\"):
ignored_lines += 1
continue
if index == ignored_lines and not is_endmarker:
comment_type = token.COMMENT # simple trailing comment
else:
comment_type = STANDALONE_COMMENT
comment = make_comment(line, preview=preview)
result.append(
ProtoComment(
type=comment_type, value=comment, newlines=nlines, consumed=consumed
)
)
nlines = 0
return result
def make_comment(content: str, *, preview: bool) -> str:
"""Return a consistently formatted comment from the given `content` string.
All comments (except for "##", "#!", "#:", '#'") should have a single
space between the hash sign and the content.
If `content` didn't start with a hash sign, one is provided.
"""
content = content.rstrip()
if not content:
return "#"
if content[0] == "#":
content = content[1:]
NON_BREAKING_SPACE = " "
if (
content
and content[0] == NON_BREAKING_SPACE
and not content.lstrip().startswith("type:")
):
content = " " + content[1:] # Replace NBSP by a simple space
if content and content[0] not in COMMENT_EXCEPTIONS[preview]:
content = " " + content
return "#" + content
def normalize_fmt_off(node: Node, *, preview: bool) -> None:
"""Convert content between `# fmt: off`/`# fmt: on` into standalone comments."""
try_again = True
while try_again:
try_again = convert_one_fmt_off_pair(node, preview=preview)
def convert_one_fmt_off_pair(node: Node, *, preview: bool) -> bool:
"""Convert content of a single `# fmt: off`/`# fmt: on` into a standalone comment.
Returns True if a pair was converted.
"""
for leaf in node.leaves():
previous_consumed = 0
for comment in list_comments(leaf.prefix, is_endmarker=False, preview=preview):
if comment.value not in FMT_PASS:
previous_consumed = comment.consumed
continue
# We only want standalone comments. If there's no previous leaf or
# the previous leaf is indentation, it's a standalone comment in
# disguise.
if comment.value in FMT_PASS and comment.type != STANDALONE_COMMENT:
prev = preceding_leaf(leaf)
if prev:
if comment.value in FMT_OFF and prev.type not in WHITESPACE:
continue
if comment.value in FMT_SKIP and prev.type in WHITESPACE:
continue
ignored_nodes = list(generate_ignored_nodes(leaf, comment, preview=preview))
if not ignored_nodes:
continue
first = ignored_nodes[0] # Can be a container node with the `leaf`.
parent = first.parent
prefix = first.prefix
if comment.value in FMT_OFF:
first.prefix = prefix[comment.consumed :]
if comment.value in FMT_SKIP:
first.prefix = ""
hidden_value = "".join(str(n) for n in ignored_nodes)
if comment.value in FMT_OFF:
hidden_value = comment.value + "\n" + hidden_value
if comment.value in FMT_SKIP:
hidden_value += " " + comment.value
if hidden_value.endswith("\n"):
# That happens when one of the `ignored_nodes` ended with a NEWLINE
# leaf (possibly followed by a DEDENT).
hidden_value = hidden_value[:-1]
first_idx: Optional[int] = None
for ignored in ignored_nodes:
index = ignored.remove()
if first_idx is None:
first_idx = index
assert parent is not None, "INTERNAL ERROR: fmt: on/off handling (1)"
assert first_idx is not None, "INTERNAL ERROR: fmt: on/off handling (2)"
parent.insert_child(
first_idx,
Leaf(
STANDALONE_COMMENT,
hidden_value,
prefix=prefix[:previous_consumed] + "\n" * comment.newlines,
),
)
return True
return False
def generate_ignored_nodes(
leaf: Leaf, comment: ProtoComment, *, preview: bool
) -> Iterator[LN]:
"""Starting from the container of `leaf`, generate all leaves until `# fmt: on`.
If comment is skip, returns leaf only.
Stops at the end of the block.
"""
container: Optional[LN] = container_of(leaf)
if comment.value in FMT_SKIP:
prev_sibling = leaf.prev_sibling
if comment.value in leaf.prefix and prev_sibling is not None:
leaf.prefix = leaf.prefix.replace(comment.value, "")
siblings = [prev_sibling]
while (
"\n" not in prev_sibling.prefix
and prev_sibling.prev_sibling is not None
):
prev_sibling = prev_sibling.prev_sibling
siblings.insert(0, prev_sibling)
for sibling in siblings:
yield sibling
elif leaf.parent is not None:
yield leaf.parent
return
while container is not None and container.type != token.ENDMARKER:
if is_fmt_on(container, preview=preview):
return
# fix for fmt: on in children
if contains_fmt_on_at_column(container, leaf.column, preview=preview):
for child in container.children:
if contains_fmt_on_at_column(child, leaf.column, preview=preview):
return
yield child
else:
yield container
container = container.next_sibling
def is_fmt_on(container: LN, preview: bool) -> bool:
"""Determine whether formatting is switched on within a container.
Determined by whether the last `# fmt:` comment is `on` or `off`.
"""
fmt_on = False
for comment in list_comments(container.prefix, is_endmarker=False, preview=preview):
if comment.value in FMT_ON:
fmt_on = True
elif comment.value in FMT_OFF:
fmt_on = False
return fmt_on
def contains_fmt_on_at_column(container: LN, column: int, *, preview: bool) -> bool:
"""Determine if children at a given column have formatting switched on."""
for child in container.children:
if (
isinstance(child, Node)
and first_leaf_column(child) == column
or isinstance(child, Leaf)
and child.column == column
):
if is_fmt_on(child, preview=preview):
return True
return False
def contains_pragma_comment(comment_list: List[Leaf]) -> bool:
"""
Returns:
True iff one of the comments in @comment_list is a pragma used by one
of the more common static analysis tools for python (e.g. mypy, flake8,
pylint).
"""
for comment in comment_list:
if comment.value.startswith(("# type:", "# noqa", "# pylint:")):
return True
return False

View file

@ -0,0 +1,57 @@
import asyncio
import logging
import sys
from typing import Any, Iterable
from black.output import err
def maybe_install_uvloop() -> None:
"""If our environment has uvloop installed we use it.
This is called only from command-line entry points to avoid
interfering with the parent process if Black is used as a library.
"""
try:
import uvloop
uvloop.install()
except ImportError:
pass
def cancel(tasks: Iterable["asyncio.Task[Any]"]) -> None:
"""asyncio signal handler that cancels all `tasks` and reports to stderr."""
err("Aborted!")
for task in tasks:
task.cancel()
def shutdown(loop: asyncio.AbstractEventLoop) -> None:
"""Cancel all pending tasks on `loop`, wait for them, and close the loop."""
try:
if sys.version_info[:2] >= (3, 7):
all_tasks = asyncio.all_tasks
else:
all_tasks = asyncio.Task.all_tasks
# This part is borrowed from asyncio/runners.py in Python 3.7b2.
to_cancel = [task for task in all_tasks(loop) if not task.done()]
if not to_cancel:
return
for task in to_cancel:
task.cancel()
if sys.version_info >= (3, 7):
loop.run_until_complete(asyncio.gather(*to_cancel, return_exceptions=True))
else:
loop.run_until_complete(
asyncio.gather(*to_cancel, loop=loop, return_exceptions=True)
)
finally:
# `concurrent.futures.Future` objects cannot be cancelled once they
# are already running. There might be some when the `shutdown()` happened.
# Silence their logger's spew about the event loop being closed.
cf_logger = logging.getLogger("concurrent.futures")
cf_logger.setLevel(logging.CRITICAL)
loop.close()

View file

@ -0,0 +1,4 @@
DEFAULT_LINE_LENGTH = 88
DEFAULT_EXCLUDES = r"/(\.direnv|\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|venv|\.svn|_build|buck-out|build|dist|__pypackages__)/" # noqa: B950
DEFAULT_INCLUDES = r"(\.pyi?|\.ipynb)$"
STDIN_PLACEHOLDER = "__BLACK_STDIN_FILENAME__"

View file

@ -0,0 +1,48 @@
from dataclasses import dataclass
from typing import Iterator, TypeVar, Union
from blib2to3.pytree import Node, Leaf, type_repr
from blib2to3.pgen2 import token
from black.nodes import Visitor
from black.output import out
from black.parsing import lib2to3_parse
LN = Union[Leaf, Node]
T = TypeVar("T")
@dataclass
class DebugVisitor(Visitor[T]):
tree_depth: int = 0
def visit_default(self, node: LN) -> Iterator[T]:
indent = " " * (2 * self.tree_depth)
if isinstance(node, Node):
_type = type_repr(node.type)
out(f"{indent}{_type}", fg="yellow")
self.tree_depth += 1
for child in node.children:
yield from self.visit(child)
self.tree_depth -= 1
out(f"{indent}/{_type}", fg="yellow", bold=False)
else:
_type = token.tok_name.get(node.type, str(node.type))
out(f"{indent}{_type}", fg="blue", nl=False)
if node.prefix:
# We don't have to handle prefixes for `Node` objects since
# that delegates to the first child anyway.
out(f" {node.prefix!r}", fg="green", bold=False, nl=False)
out(f" {node.value!r}", fg="blue", bold=False)
@classmethod
def show(cls, code: Union[str, Leaf, Node]) -> None:
"""Pretty-print the lib2to3 AST of a given string of `code`.
Convenience method for debugging.
"""
v: DebugVisitor[None] = DebugVisitor()
if isinstance(code, str):
code = lib2to3_parse(code)
list(v.visit(code))

View file

@ -0,0 +1,278 @@
from functools import lru_cache
import io
import os
from pathlib import Path
import sys
from typing import (
Any,
Dict,
Iterable,
Iterator,
List,
Optional,
Pattern,
Sequence,
Tuple,
Union,
TYPE_CHECKING,
)
from mypy_extensions import mypyc_attr
from pathspec import PathSpec
from pathspec.patterns.gitwildmatch import GitWildMatchPatternError
if sys.version_info >= (3, 11):
import tomllib
else:
import tomli as tomllib
from black.output import err
from black.report import Report
from black.handle_ipynb_magics import jupyter_dependencies_are_installed
if TYPE_CHECKING:
import colorama # noqa: F401
@lru_cache()
def find_project_root(srcs: Sequence[str]) -> Tuple[Path, str]:
"""Return a directory containing .git, .hg, or pyproject.toml.
That directory will be a common parent of all files and directories
passed in `srcs`.
If no directory in the tree contains a marker that would specify it's the
project root, the root of the file system is returned.
Returns a two-tuple with the first element as the project root path and
the second element as a string describing the method by which the
project root was discovered.
"""
if not srcs:
srcs = [str(Path.cwd().resolve())]
path_srcs = [Path(Path.cwd(), src).resolve() for src in srcs]
# A list of lists of parents for each 'src'. 'src' is included as a
# "parent" of itself if it is a directory
src_parents = [
list(path.parents) + ([path] if path.is_dir() else []) for path in path_srcs
]
common_base = max(
set.intersection(*(set(parents) for parents in src_parents)),
key=lambda path: path.parts,
)
for directory in (common_base, *common_base.parents):
if (directory / ".git").exists():
return directory, ".git directory"
if (directory / ".hg").is_dir():
return directory, ".hg directory"
if (directory / "pyproject.toml").is_file():
return directory, "pyproject.toml"
return directory, "file system root"
def find_pyproject_toml(path_search_start: Tuple[str, ...]) -> Optional[str]:
"""Find the absolute filepath to a pyproject.toml if it exists"""
path_project_root, _ = find_project_root(path_search_start)
path_pyproject_toml = path_project_root / "pyproject.toml"
if path_pyproject_toml.is_file():
return str(path_pyproject_toml)
try:
path_user_pyproject_toml = find_user_pyproject_toml()
return (
str(path_user_pyproject_toml)
if path_user_pyproject_toml.is_file()
else None
)
except (PermissionError, RuntimeError) as e:
# We do not have access to the user-level config directory, so ignore it.
err(f"Ignoring user configuration directory due to {e!r}")
return None
@mypyc_attr(patchable=True)
def parse_pyproject_toml(path_config: str) -> Dict[str, Any]:
"""Parse a pyproject toml file, pulling out relevant parts for Black
If parsing fails, will raise a tomllib.TOMLDecodeError
"""
with open(path_config, "rb") as f:
pyproject_toml = tomllib.load(f)
config = pyproject_toml.get("tool", {}).get("black", {})
return {k.replace("--", "").replace("-", "_"): v for k, v in config.items()}
@lru_cache()
def find_user_pyproject_toml() -> Path:
r"""Return the path to the top-level user configuration for black.
This looks for ~\.black on Windows and ~/.config/black on Linux and other
Unix systems.
May raise:
- RuntimeError: if the current user has no homedir
- PermissionError: if the current process cannot access the user's homedir
"""
if sys.platform == "win32":
# Windows
user_config_path = Path.home() / ".black"
else:
config_root = os.environ.get("XDG_CONFIG_HOME", "~/.config")
user_config_path = Path(config_root).expanduser() / "black"
return user_config_path.resolve()
@lru_cache()
def get_gitignore(root: Path) -> PathSpec:
"""Return a PathSpec matching gitignore content if present."""
gitignore = root / ".gitignore"
lines: List[str] = []
if gitignore.is_file():
with gitignore.open(encoding="utf-8") as gf:
lines = gf.readlines()
try:
return PathSpec.from_lines("gitwildmatch", lines)
except GitWildMatchPatternError as e:
err(f"Could not parse {gitignore}: {e}")
raise
def normalize_path_maybe_ignore(
path: Path,
root: Path,
report: Optional[Report] = None,
) -> Optional[str]:
"""Normalize `path`. May return `None` if `path` was ignored.
`report` is where "path ignored" output goes.
"""
try:
abspath = path if path.is_absolute() else Path.cwd() / path
normalized_path = abspath.resolve()
try:
root_relative_path = normalized_path.relative_to(root).as_posix()
except ValueError:
if report:
report.path_ignored(
path, f"is a symbolic link that points outside {root}"
)
return None
except OSError as e:
if report:
report.path_ignored(path, f"cannot be read because {e}")
return None
return root_relative_path
def path_is_excluded(
normalized_path: str,
pattern: Optional[Pattern[str]],
) -> bool:
match = pattern.search(normalized_path) if pattern else None
return bool(match and match.group(0))
def gen_python_files(
paths: Iterable[Path],
root: Path,
include: Pattern[str],
exclude: Pattern[str],
extend_exclude: Optional[Pattern[str]],
force_exclude: Optional[Pattern[str]],
report: Report,
gitignore: Optional[PathSpec],
*,
verbose: bool,
quiet: bool,
) -> Iterator[Path]:
"""Generate all files under `path` whose paths are not excluded by the
`exclude_regex`, `extend_exclude`, or `force_exclude` regexes,
but are included by the `include` regex.
Symbolic links pointing outside of the `root` directory are ignored.
`report` is where output about exclusions goes.
"""
assert root.is_absolute(), f"INTERNAL ERROR: `root` must be absolute but is {root}"
for child in paths:
normalized_path = normalize_path_maybe_ignore(child, root, report)
if normalized_path is None:
continue
# First ignore files matching .gitignore, if passed
if gitignore is not None and gitignore.match_file(normalized_path):
report.path_ignored(child, "matches the .gitignore file content")
continue
# Then ignore with `--exclude` `--extend-exclude` and `--force-exclude` options.
normalized_path = "/" + normalized_path
if child.is_dir():
normalized_path += "/"
if path_is_excluded(normalized_path, exclude):
report.path_ignored(child, "matches the --exclude regular expression")
continue
if path_is_excluded(normalized_path, extend_exclude):
report.path_ignored(
child, "matches the --extend-exclude regular expression"
)
continue
if path_is_excluded(normalized_path, force_exclude):
report.path_ignored(child, "matches the --force-exclude regular expression")
continue
if child.is_dir():
# If gitignore is None, gitignore usage is disabled, while a Falsey
# gitignore is when the directory doesn't have a .gitignore file.
yield from gen_python_files(
child.iterdir(),
root,
include,
exclude,
extend_exclude,
force_exclude,
report,
gitignore + get_gitignore(child) if gitignore is not None else None,
verbose=verbose,
quiet=quiet,
)
elif child.is_file():
if child.suffix == ".ipynb" and not jupyter_dependencies_are_installed(
verbose=verbose, quiet=quiet
):
continue
include_match = include.search(normalized_path) if include else True
if include_match:
yield child
def wrap_stream_for_windows(
f: io.TextIOWrapper,
) -> Union[io.TextIOWrapper, "colorama.AnsiToWin32"]:
"""
Wrap stream with colorama's wrap_stream so colors are shown on Windows.
If `colorama` is unavailable, the original stream is returned unmodified.
Otherwise, the `wrap_stream()` function determines whether the stream needs
to be wrapped for a Windows environment and will accordingly either return
an `AnsiToWin32` wrapper or the original stream.
"""
try:
from colorama.initialise import wrap_stream
except ImportError:
return f
else:
# Set `strip=False` to avoid needing to modify test_express_diff_with_color.
return wrap_stream(f, convert=None, strip=False, autoreset=False, wrap=True)

View file

@ -0,0 +1,465 @@
"""Functions to process IPython magics with."""
from functools import lru_cache
import dataclasses
import ast
from typing import Dict, List, Tuple, Optional
import secrets
import sys
import collections
if sys.version_info >= (3, 10):
from typing import TypeGuard
else:
from typing_extensions import TypeGuard
from black.report import NothingChanged
from black.output import out
TRANSFORMED_MAGICS = frozenset(
(
"get_ipython().run_cell_magic",
"get_ipython().system",
"get_ipython().getoutput",
"get_ipython().run_line_magic",
)
)
TOKENS_TO_IGNORE = frozenset(
(
"ENDMARKER",
"NL",
"NEWLINE",
"COMMENT",
"DEDENT",
"UNIMPORTANT_WS",
"ESCAPED_NL",
)
)
PYTHON_CELL_MAGICS = frozenset(
(
"capture",
"prun",
"pypy",
"python",
"python3",
"time",
"timeit",
)
)
TOKEN_HEX = secrets.token_hex
@dataclasses.dataclass(frozen=True)
class Replacement:
mask: str
src: str
@lru_cache()
def jupyter_dependencies_are_installed(*, verbose: bool, quiet: bool) -> bool:
try:
import IPython # noqa:F401
import tokenize_rt # noqa:F401
except ModuleNotFoundError:
if verbose or not quiet:
msg = (
"Skipping .ipynb files as Jupyter dependencies are not installed.\n"
"You can fix this by running ``pip install black[jupyter]``"
)
out(msg)
return False
else:
return True
def remove_trailing_semicolon(src: str) -> Tuple[str, bool]:
"""Remove trailing semicolon from Jupyter notebook cell.
For example,
fig, ax = plt.subplots()
ax.plot(x_data, y_data); # plot data
would become
fig, ax = plt.subplots()
ax.plot(x_data, y_data) # plot data
Mirrors the logic in `quiet` from `IPython.core.displayhook`, but uses
``tokenize_rt`` so that round-tripping works fine.
"""
from tokenize_rt import (
src_to_tokens,
tokens_to_src,
reversed_enumerate,
)
tokens = src_to_tokens(src)
trailing_semicolon = False
for idx, token in reversed_enumerate(tokens):
if token.name in TOKENS_TO_IGNORE:
continue
if token.name == "OP" and token.src == ";":
del tokens[idx]
trailing_semicolon = True
break
if not trailing_semicolon:
return src, False
return tokens_to_src(tokens), True
def put_trailing_semicolon_back(src: str, has_trailing_semicolon: bool) -> str:
"""Put trailing semicolon back if cell originally had it.
Mirrors the logic in `quiet` from `IPython.core.displayhook`, but uses
``tokenize_rt`` so that round-tripping works fine.
"""
if not has_trailing_semicolon:
return src
from tokenize_rt import src_to_tokens, tokens_to_src, reversed_enumerate
tokens = src_to_tokens(src)
for idx, token in reversed_enumerate(tokens):
if token.name in TOKENS_TO_IGNORE:
continue
tokens[idx] = token._replace(src=token.src + ";")
break
else: # pragma: nocover
raise AssertionError(
"INTERNAL ERROR: Was not able to reinstate trailing semicolon. "
"Please report a bug on https://github.com/psf/black/issues. "
) from None
return str(tokens_to_src(tokens))
def mask_cell(src: str) -> Tuple[str, List[Replacement]]:
"""Mask IPython magics so content becomes parseable Python code.
For example,
%matplotlib inline
'foo'
becomes
"25716f358c32750e"
'foo'
The replacements are returned, along with the transformed code.
"""
replacements: List[Replacement] = []
try:
ast.parse(src)
except SyntaxError:
# Might have IPython magics, will process below.
pass
else:
# Syntax is fine, nothing to mask, early return.
return src, replacements
from IPython.core.inputtransformer2 import TransformerManager
transformer_manager = TransformerManager()
transformed = transformer_manager.transform_cell(src)
transformed, cell_magic_replacements = replace_cell_magics(transformed)
replacements += cell_magic_replacements
transformed = transformer_manager.transform_cell(transformed)
transformed, magic_replacements = replace_magics(transformed)
if len(transformed.splitlines()) != len(src.splitlines()):
# Multi-line magic, not supported.
raise NothingChanged
replacements += magic_replacements
return transformed, replacements
def get_token(src: str, magic: str) -> str:
"""Return randomly generated token to mask IPython magic with.
For example, if 'magic' was `%matplotlib inline`, then a possible
token to mask it with would be `"43fdd17f7e5ddc83"`. The token
will be the same length as the magic, and we make sure that it was
not already present anywhere else in the cell.
"""
assert magic
nbytes = max(len(magic) // 2 - 1, 1)
token = TOKEN_HEX(nbytes)
counter = 0
while token in src:
token = TOKEN_HEX(nbytes)
counter += 1
if counter > 100:
raise AssertionError(
"INTERNAL ERROR: Black was not able to replace IPython magic. "
"Please report a bug on https://github.com/psf/black/issues. "
f"The magic might be helpful: {magic}"
) from None
if len(token) + 2 < len(magic):
token = f"{token}."
return f'"{token}"'
def replace_cell_magics(src: str) -> Tuple[str, List[Replacement]]:
"""Replace cell magic with token.
Note that 'src' will already have been processed by IPython's
TransformerManager().transform_cell.
Example,
get_ipython().run_cell_magic('t', '-n1', 'ls =!ls\\n')
becomes
"a794."
ls =!ls
The replacement, along with the transformed code, is returned.
"""
replacements: List[Replacement] = []
tree = ast.parse(src)
cell_magic_finder = CellMagicFinder()
cell_magic_finder.visit(tree)
if cell_magic_finder.cell_magic is None:
return src, replacements
header = cell_magic_finder.cell_magic.header
mask = get_token(src, header)
replacements.append(Replacement(mask=mask, src=header))
return f"{mask}\n{cell_magic_finder.cell_magic.body}", replacements
def replace_magics(src: str) -> Tuple[str, List[Replacement]]:
"""Replace magics within body of cell.
Note that 'src' will already have been processed by IPython's
TransformerManager().transform_cell.
Example, this
get_ipython().run_line_magic('matplotlib', 'inline')
'foo'
becomes
"5e67db56d490fd39"
'foo'
The replacement, along with the transformed code, are returned.
"""
replacements = []
magic_finder = MagicFinder()
magic_finder.visit(ast.parse(src))
new_srcs = []
for i, line in enumerate(src.splitlines(), start=1):
if i in magic_finder.magics:
offsets_and_magics = magic_finder.magics[i]
if len(offsets_and_magics) != 1: # pragma: nocover
raise AssertionError(
f"Expecting one magic per line, got: {offsets_and_magics}\n"
"Please report a bug on https://github.com/psf/black/issues."
)
col_offset, magic = (
offsets_and_magics[0].col_offset,
offsets_and_magics[0].magic,
)
mask = get_token(src, magic)
replacements.append(Replacement(mask=mask, src=magic))
line = line[:col_offset] + mask
new_srcs.append(line)
return "\n".join(new_srcs), replacements
def unmask_cell(src: str, replacements: List[Replacement]) -> str:
"""Remove replacements from cell.
For example
"9b20"
foo = bar
becomes
%%time
foo = bar
"""
for replacement in replacements:
src = src.replace(replacement.mask, replacement.src)
return src
def _is_ipython_magic(node: ast.expr) -> TypeGuard[ast.Attribute]:
"""Check if attribute is IPython magic.
Note that the source of the abstract syntax tree
will already have been processed by IPython's
TransformerManager().transform_cell.
"""
return (
isinstance(node, ast.Attribute)
and isinstance(node.value, ast.Call)
and isinstance(node.value.func, ast.Name)
and node.value.func.id == "get_ipython"
)
def _get_str_args(args: List[ast.expr]) -> List[str]:
str_args = []
for arg in args:
assert isinstance(arg, ast.Str)
str_args.append(arg.s)
return str_args
@dataclasses.dataclass(frozen=True)
class CellMagic:
name: str
params: Optional[str]
body: str
@property
def header(self) -> str:
if self.params:
return f"%%{self.name} {self.params}"
return f"%%{self.name}"
# ast.NodeVisitor + dataclass = breakage under mypyc.
class CellMagicFinder(ast.NodeVisitor):
"""Find cell magics.
Note that the source of the abstract syntax tree
will already have been processed by IPython's
TransformerManager().transform_cell.
For example,
%%time\nfoo()
would have been transformed to
get_ipython().run_cell_magic('time', '', 'foo()\\n')
and we look for instances of the latter.
"""
def __init__(self, cell_magic: Optional[CellMagic] = None) -> None:
self.cell_magic = cell_magic
def visit_Expr(self, node: ast.Expr) -> None:
"""Find cell magic, extract header and body."""
if (
isinstance(node.value, ast.Call)
and _is_ipython_magic(node.value.func)
and node.value.func.attr == "run_cell_magic"
):
args = _get_str_args(node.value.args)
self.cell_magic = CellMagic(name=args[0], params=args[1], body=args[2])
self.generic_visit(node)
@dataclasses.dataclass(frozen=True)
class OffsetAndMagic:
col_offset: int
magic: str
# Unsurprisingly, subclassing ast.NodeVisitor means we can't use dataclasses here
# as mypyc will generate broken code.
class MagicFinder(ast.NodeVisitor):
"""Visit cell to look for get_ipython calls.
Note that the source of the abstract syntax tree
will already have been processed by IPython's
TransformerManager().transform_cell.
For example,
%matplotlib inline
would have been transformed to
get_ipython().run_line_magic('matplotlib', 'inline')
and we look for instances of the latter (and likewise for other
types of magics).
"""
def __init__(self) -> None:
self.magics: Dict[int, List[OffsetAndMagic]] = collections.defaultdict(list)
def visit_Assign(self, node: ast.Assign) -> None:
"""Look for system assign magics.
For example,
black_version = !black --version
env = %env var
would have been (respectively) transformed to
black_version = get_ipython().getoutput('black --version')
env = get_ipython().run_line_magic('env', 'var')
and we look for instances of any of the latter.
"""
if isinstance(node.value, ast.Call) and _is_ipython_magic(node.value.func):
args = _get_str_args(node.value.args)
if node.value.func.attr == "getoutput":
src = f"!{args[0]}"
elif node.value.func.attr == "run_line_magic":
src = f"%{args[0]}"
if args[1]:
src += f" {args[1]}"
else:
raise AssertionError(
f"Unexpected IPython magic {node.value.func.attr!r} found. "
"Please report a bug on https://github.com/psf/black/issues."
) from None
self.magics[node.value.lineno].append(
OffsetAndMagic(node.value.col_offset, src)
)
self.generic_visit(node)
def visit_Expr(self, node: ast.Expr) -> None:
"""Look for magics in body of cell.
For examples,
!ls
!!ls
?ls
??ls
would (respectively) get transformed to
get_ipython().system('ls')
get_ipython().getoutput('ls')
get_ipython().run_line_magic('pinfo', 'ls')
get_ipython().run_line_magic('pinfo2', 'ls')
and we look for instances of any of the latter.
"""
if isinstance(node.value, ast.Call) and _is_ipython_magic(node.value.func):
args = _get_str_args(node.value.args)
if node.value.func.attr == "run_line_magic":
if args[0] == "pinfo":
src = f"?{args[1]}"
elif args[0] == "pinfo2":
src = f"??{args[1]}"
else:
src = f"%{args[0]}"
if args[1]:
src += f" {args[1]}"
elif node.value.func.attr == "system":
src = f"!{args[0]}"
elif node.value.func.attr == "getoutput":
src = f"!!{args[0]}"
else:
raise NothingChanged # unsupported magic.
self.magics[node.value.lineno].append(
OffsetAndMagic(node.value.col_offset, src)
)
self.generic_visit(node)

File diff suppressed because it is too large Load diff

View file

@ -0,0 +1,773 @@
from dataclasses import dataclass, field
import itertools
import sys
from typing import (
Callable,
Dict,
Iterator,
List,
Optional,
Sequence,
Tuple,
TypeVar,
cast,
)
from blib2to3.pytree import Node, Leaf
from blib2to3.pgen2 import token
from black.brackets import BracketTracker, DOT_PRIORITY
from black.mode import Mode, Preview
from black.nodes import STANDALONE_COMMENT, TEST_DESCENDANTS
from black.nodes import BRACKETS, OPENING_BRACKETS, CLOSING_BRACKETS
from black.nodes import syms, whitespace, replace_child, child_towards
from black.nodes import is_multiline_string, is_import, is_type_comment
from black.nodes import is_one_sequence_between
# types
T = TypeVar("T")
Index = int
LeafID = int
@dataclass
class Line:
"""Holds leaves and comments. Can be printed with `str(line)`."""
mode: Mode
depth: int = 0
leaves: List[Leaf] = field(default_factory=list)
# keys ordered like `leaves`
comments: Dict[LeafID, List[Leaf]] = field(default_factory=dict)
bracket_tracker: BracketTracker = field(default_factory=BracketTracker)
inside_brackets: bool = False
should_split_rhs: bool = False
magic_trailing_comma: Optional[Leaf] = None
def append(self, leaf: Leaf, preformatted: bool = False) -> None:
"""Add a new `leaf` to the end of the line.
Unless `preformatted` is True, the `leaf` will receive a new consistent
whitespace prefix and metadata applied by :class:`BracketTracker`.
Trailing commas are maybe removed, unpacked for loop variables are
demoted from being delimiters.
Inline comments are put aside.
"""
has_value = leaf.type in BRACKETS or bool(leaf.value.strip())
if not has_value:
return
if token.COLON == leaf.type and self.is_class_paren_empty:
del self.leaves[-2:]
if self.leaves and not preformatted:
# Note: at this point leaf.prefix should be empty except for
# imports, for which we only preserve newlines.
leaf.prefix += whitespace(
leaf, complex_subscript=self.is_complex_subscript(leaf)
)
if self.inside_brackets or not preformatted:
self.bracket_tracker.mark(leaf)
if self.mode.magic_trailing_comma:
if self.has_magic_trailing_comma(leaf):
self.magic_trailing_comma = leaf
elif self.has_magic_trailing_comma(leaf, ensure_removable=True):
self.remove_trailing_comma()
if not self.append_comment(leaf):
self.leaves.append(leaf)
def append_safe(self, leaf: Leaf, preformatted: bool = False) -> None:
"""Like :func:`append()` but disallow invalid standalone comment structure.
Raises ValueError when any `leaf` is appended after a standalone comment
or when a standalone comment is not the first leaf on the line.
"""
if self.bracket_tracker.depth == 0:
if self.is_comment:
raise ValueError("cannot append to standalone comments")
if self.leaves and leaf.type == STANDALONE_COMMENT:
raise ValueError(
"cannot append standalone comments to a populated line"
)
self.append(leaf, preformatted=preformatted)
@property
def is_comment(self) -> bool:
"""Is this line a standalone comment?"""
return len(self.leaves) == 1 and self.leaves[0].type == STANDALONE_COMMENT
@property
def is_decorator(self) -> bool:
"""Is this line a decorator?"""
return bool(self) and self.leaves[0].type == token.AT
@property
def is_import(self) -> bool:
"""Is this an import line?"""
return bool(self) and is_import(self.leaves[0])
@property
def is_class(self) -> bool:
"""Is this line a class definition?"""
return (
bool(self)
and self.leaves[0].type == token.NAME
and self.leaves[0].value == "class"
)
@property
def is_stub_class(self) -> bool:
"""Is this line a class definition with a body consisting only of "..."?"""
return self.is_class and self.leaves[-3:] == [
Leaf(token.DOT, ".") for _ in range(3)
]
@property
def is_def(self) -> bool:
"""Is this a function definition? (Also returns True for async defs.)"""
try:
first_leaf = self.leaves[0]
except IndexError:
return False
try:
second_leaf: Optional[Leaf] = self.leaves[1]
except IndexError:
second_leaf = None
return (first_leaf.type == token.NAME and first_leaf.value == "def") or (
first_leaf.type == token.ASYNC
and second_leaf is not None
and second_leaf.type == token.NAME
and second_leaf.value == "def"
)
@property
def is_class_paren_empty(self) -> bool:
"""Is this a class with no base classes but using parentheses?
Those are unnecessary and should be removed.
"""
return (
bool(self)
and len(self.leaves) == 4
and self.is_class
and self.leaves[2].type == token.LPAR
and self.leaves[2].value == "("
and self.leaves[3].type == token.RPAR
and self.leaves[3].value == ")"
)
@property
def is_triple_quoted_string(self) -> bool:
"""Is the line a triple quoted string?"""
return (
bool(self)
and self.leaves[0].type == token.STRING
and self.leaves[0].value.startswith(('"""', "'''"))
)
def contains_standalone_comments(self, depth_limit: int = sys.maxsize) -> bool:
"""If so, needs to be split before emitting."""
for leaf in self.leaves:
if leaf.type == STANDALONE_COMMENT and leaf.bracket_depth <= depth_limit:
return True
return False
def contains_uncollapsable_type_comments(self) -> bool:
ignored_ids = set()
try:
last_leaf = self.leaves[-1]
ignored_ids.add(id(last_leaf))
if last_leaf.type == token.COMMA or (
last_leaf.type == token.RPAR and not last_leaf.value
):
# When trailing commas or optional parens are inserted by Black for
# consistency, comments after the previous last element are not moved
# (they don't have to, rendering will still be correct). So we ignore
# trailing commas and invisible.
last_leaf = self.leaves[-2]
ignored_ids.add(id(last_leaf))
except IndexError:
return False
# A type comment is uncollapsable if it is attached to a leaf
# that isn't at the end of the line (since that could cause it
# to get associated to a different argument) or if there are
# comments before it (since that could cause it to get hidden
# behind a comment.
comment_seen = False
for leaf_id, comments in self.comments.items():
for comment in comments:
if is_type_comment(comment):
if comment_seen or (
not is_type_comment(comment, " ignore")
and leaf_id not in ignored_ids
):
return True
comment_seen = True
return False
def contains_unsplittable_type_ignore(self) -> bool:
if not self.leaves:
return False
# If a 'type: ignore' is attached to the end of a line, we
# can't split the line, because we can't know which of the
# subexpressions the ignore was meant to apply to.
#
# We only want this to apply to actual physical lines from the
# original source, though: we don't want the presence of a
# 'type: ignore' at the end of a multiline expression to
# justify pushing it all onto one line. Thus we
# (unfortunately) need to check the actual source lines and
# only report an unsplittable 'type: ignore' if this line was
# one line in the original code.
# Grab the first and last line numbers, skipping generated leaves
first_line = next((leaf.lineno for leaf in self.leaves if leaf.lineno != 0), 0)
last_line = next(
(leaf.lineno for leaf in reversed(self.leaves) if leaf.lineno != 0), 0
)
if first_line == last_line:
# We look at the last two leaves since a comma or an
# invisible paren could have been added at the end of the
# line.
for node in self.leaves[-2:]:
for comment in self.comments.get(id(node), []):
if is_type_comment(comment, " ignore"):
return True
return False
def contains_multiline_strings(self) -> bool:
return any(is_multiline_string(leaf) for leaf in self.leaves)
def has_magic_trailing_comma(
self, closing: Leaf, ensure_removable: bool = False
) -> bool:
"""Return True if we have a magic trailing comma, that is when:
- there's a trailing comma here
- it's not a one-tuple
- it's not a single-element subscript
Additionally, if ensure_removable:
- it's not from square bracket indexing
"""
if not (
closing.type in CLOSING_BRACKETS
and self.leaves
and self.leaves[-1].type == token.COMMA
):
return False
if closing.type == token.RBRACE:
return True
if closing.type == token.RSQB:
if (
Preview.one_element_subscript in self.mode
and closing.parent
and closing.parent.type == syms.trailer
and closing.opening_bracket
and is_one_sequence_between(
closing.opening_bracket,
closing,
self.leaves,
brackets=(token.LSQB, token.RSQB),
)
):
return False
if not ensure_removable:
return True
comma = self.leaves[-1]
return bool(comma.parent and comma.parent.type == syms.listmaker)
if self.is_import:
return True
if closing.opening_bracket is not None and not is_one_sequence_between(
closing.opening_bracket, closing, self.leaves
):
return True
return False
def append_comment(self, comment: Leaf) -> bool:
"""Add an inline or standalone comment to the line."""
if (
comment.type == STANDALONE_COMMENT
and self.bracket_tracker.any_open_brackets()
):
comment.prefix = ""
return False
if comment.type != token.COMMENT:
return False
if not self.leaves:
comment.type = STANDALONE_COMMENT
comment.prefix = ""
return False
last_leaf = self.leaves[-1]
if (
last_leaf.type == token.RPAR
and not last_leaf.value
and last_leaf.parent
and len(list(last_leaf.parent.leaves())) <= 3
and not is_type_comment(comment)
):
# Comments on an optional parens wrapping a single leaf should belong to
# the wrapped node except if it's a type comment. Pinning the comment like
# this avoids unstable formatting caused by comment migration.
if len(self.leaves) < 2:
comment.type = STANDALONE_COMMENT
comment.prefix = ""
return False
last_leaf = self.leaves[-2]
self.comments.setdefault(id(last_leaf), []).append(comment)
return True
def comments_after(self, leaf: Leaf) -> List[Leaf]:
"""Generate comments that should appear directly after `leaf`."""
return self.comments.get(id(leaf), [])
def remove_trailing_comma(self) -> None:
"""Remove the trailing comma and moves the comments attached to it."""
trailing_comma = self.leaves.pop()
trailing_comma_comments = self.comments.pop(id(trailing_comma), [])
self.comments.setdefault(id(self.leaves[-1]), []).extend(
trailing_comma_comments
)
def is_complex_subscript(self, leaf: Leaf) -> bool:
"""Return True iff `leaf` is part of a slice with non-trivial exprs."""
open_lsqb = self.bracket_tracker.get_open_lsqb()
if open_lsqb is None:
return False
subscript_start = open_lsqb.next_sibling
if isinstance(subscript_start, Node):
if subscript_start.type == syms.listmaker:
return False
if subscript_start.type == syms.subscriptlist:
subscript_start = child_towards(subscript_start, leaf)
return subscript_start is not None and any(
n.type in TEST_DESCENDANTS for n in subscript_start.pre_order()
)
def enumerate_with_length(
self, reversed: bool = False
) -> Iterator[Tuple[Index, Leaf, int]]:
"""Return an enumeration of leaves with their length.
Stops prematurely on multiline strings and standalone comments.
"""
op = cast(
Callable[[Sequence[Leaf]], Iterator[Tuple[Index, Leaf]]],
enumerate_reversed if reversed else enumerate,
)
for index, leaf in op(self.leaves):
length = len(leaf.prefix) + len(leaf.value)
if "\n" in leaf.value:
return # Multiline strings, we can't continue.
for comment in self.comments_after(leaf):
length += len(comment.value)
yield index, leaf, length
def clone(self) -> "Line":
return Line(
mode=self.mode,
depth=self.depth,
inside_brackets=self.inside_brackets,
should_split_rhs=self.should_split_rhs,
magic_trailing_comma=self.magic_trailing_comma,
)
def __str__(self) -> str:
"""Render the line."""
if not self:
return "\n"
indent = " " * self.depth
leaves = iter(self.leaves)
first = next(leaves)
res = f"{first.prefix}{indent}{first.value}"
for leaf in leaves:
res += str(leaf)
for comment in itertools.chain.from_iterable(self.comments.values()):
res += str(comment)
return res + "\n"
def __bool__(self) -> bool:
"""Return True if the line has leaves or comments."""
return bool(self.leaves or self.comments)
@dataclass
class EmptyLineTracker:
"""Provides a stateful method that returns the number of potential extra
empty lines needed before and after the currently processed line.
Note: this tracker works on lines that haven't been split yet. It assumes
the prefix of the first leaf consists of optional newlines. Those newlines
are consumed by `maybe_empty_lines()` and included in the computation.
"""
is_pyi: bool = False
previous_line: Optional[Line] = None
previous_after: int = 0
previous_defs: List[int] = field(default_factory=list)
def maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]:
"""Return the number of extra empty lines before and after the `current_line`.
This is for separating `def`, `async def` and `class` with extra empty
lines (two on module-level).
"""
before, after = self._maybe_empty_lines(current_line)
before = (
# Black should not insert empty lines at the beginning
# of the file
0
if self.previous_line is None
else before - self.previous_after
)
self.previous_after = after
self.previous_line = current_line
return before, after
def _maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]:
max_allowed = 1
if current_line.depth == 0:
max_allowed = 1 if self.is_pyi else 2
if current_line.leaves:
# Consume the first leaf's extra newlines.
first_leaf = current_line.leaves[0]
before = first_leaf.prefix.count("\n")
before = min(before, max_allowed)
first_leaf.prefix = ""
else:
before = 0
depth = current_line.depth
while self.previous_defs and self.previous_defs[-1] >= depth:
if self.is_pyi:
assert self.previous_line is not None
if depth and not current_line.is_def and self.previous_line.is_def:
# Empty lines between attributes and methods should be preserved.
before = min(1, before)
elif depth:
before = 0
else:
before = 1
else:
if depth:
before = 1
elif (
not depth
and self.previous_defs[-1]
and current_line.leaves[-1].type == token.COLON
and (
current_line.leaves[0].value
not in ("with", "try", "for", "while", "if", "match")
)
):
# We shouldn't add two newlines between an indented function and
# a dependent non-indented clause. This is to avoid issues with
# conditional function definitions that are technically top-level
# and therefore get two trailing newlines, but look weird and
# inconsistent when they're followed by elif, else, etc. This is
# worse because these functions only get *one* preceding newline
# already.
before = 1
else:
before = 2
self.previous_defs.pop()
if current_line.is_decorator or current_line.is_def or current_line.is_class:
return self._maybe_empty_lines_for_class_or_def(current_line, before)
if (
self.previous_line
and self.previous_line.is_import
and not current_line.is_import
and depth == self.previous_line.depth
):
return (before or 1), 0
if (
self.previous_line
and self.previous_line.is_class
and current_line.is_triple_quoted_string
):
return before, 1
return before, 0
def _maybe_empty_lines_for_class_or_def(
self, current_line: Line, before: int
) -> Tuple[int, int]:
if not current_line.is_decorator:
self.previous_defs.append(current_line.depth)
if self.previous_line is None:
# Don't insert empty lines before the first line in the file.
return 0, 0
if self.previous_line.is_decorator:
if self.is_pyi and current_line.is_stub_class:
# Insert an empty line after a decorated stub class
return 0, 1
return 0, 0
if self.previous_line.depth < current_line.depth and (
self.previous_line.is_class or self.previous_line.is_def
):
return 0, 0
if (
self.previous_line.is_comment
and self.previous_line.depth == current_line.depth
and before == 0
):
return 0, 0
if self.is_pyi:
if current_line.is_class or self.previous_line.is_class:
if self.previous_line.depth < current_line.depth:
newlines = 0
elif self.previous_line.depth > current_line.depth:
newlines = 1
elif current_line.is_stub_class and self.previous_line.is_stub_class:
# No blank line between classes with an empty body
newlines = 0
else:
newlines = 1
elif (
current_line.is_def or current_line.is_decorator
) and not self.previous_line.is_def:
if current_line.depth:
# In classes empty lines between attributes and methods should
# be preserved.
newlines = min(1, before)
else:
# Blank line between a block of functions (maybe with preceding
# decorators) and a block of non-functions
newlines = 1
elif self.previous_line.depth > current_line.depth:
newlines = 1
else:
newlines = 0
else:
newlines = 1 if current_line.depth else 2
return newlines, 0
def enumerate_reversed(sequence: Sequence[T]) -> Iterator[Tuple[Index, T]]:
"""Like `reversed(enumerate(sequence))` if that were possible."""
index = len(sequence) - 1
for element in reversed(sequence):
yield (index, element)
index -= 1
def append_leaves(
new_line: Line, old_line: Line, leaves: List[Leaf], preformatted: bool = False
) -> None:
"""
Append leaves (taken from @old_line) to @new_line, making sure to fix the
underlying Node structure where appropriate.
All of the leaves in @leaves are duplicated. The duplicates are then
appended to @new_line and used to replace their originals in the underlying
Node structure. Any comments attached to the old leaves are reattached to
the new leaves.
Pre-conditions:
set(@leaves) is a subset of set(@old_line.leaves).
"""
for old_leaf in leaves:
new_leaf = Leaf(old_leaf.type, old_leaf.value)
replace_child(old_leaf, new_leaf)
new_line.append(new_leaf, preformatted=preformatted)
for comment_leaf in old_line.comments_after(old_leaf):
new_line.append(comment_leaf, preformatted=True)
def is_line_short_enough(line: Line, *, line_length: int, line_str: str = "") -> bool:
"""Return True if `line` is no longer than `line_length`.
Uses the provided `line_str` rendering, if any, otherwise computes a new one.
"""
if not line_str:
line_str = line_to_string(line)
return (
len(line_str) <= line_length
and "\n" not in line_str # multiline strings
and not line.contains_standalone_comments()
)
def can_be_split(line: Line) -> bool:
"""Return False if the line cannot be split *for sure*.
This is not an exhaustive search but a cheap heuristic that we can use to
avoid some unfortunate formattings (mostly around wrapping unsplittable code
in unnecessary parentheses).
"""
leaves = line.leaves
if len(leaves) < 2:
return False
if leaves[0].type == token.STRING and leaves[1].type == token.DOT:
call_count = 0
dot_count = 0
next = leaves[-1]
for leaf in leaves[-2::-1]:
if leaf.type in OPENING_BRACKETS:
if next.type not in CLOSING_BRACKETS:
return False
call_count += 1
elif leaf.type == token.DOT:
dot_count += 1
elif leaf.type == token.NAME:
if not (next.type == token.DOT or next.type in OPENING_BRACKETS):
return False
elif leaf.type not in CLOSING_BRACKETS:
return False
if dot_count > 1 and call_count > 1:
return False
return True
def can_omit_invisible_parens(
line: Line,
line_length: int,
) -> bool:
"""Does `line` have a shape safe to reformat without optional parens around it?
Returns True for only a subset of potentially nice looking formattings but
the point is to not return false positives that end up producing lines that
are too long.
"""
bt = line.bracket_tracker
if not bt.delimiters:
# Without delimiters the optional parentheses are useless.
return True
max_priority = bt.max_delimiter_priority()
if bt.delimiter_count_with_priority(max_priority) > 1:
# With more than one delimiter of a kind the optional parentheses read better.
return False
if max_priority == DOT_PRIORITY:
# A single stranded method call doesn't require optional parentheses.
return True
assert len(line.leaves) >= 2, "Stranded delimiter"
# With a single delimiter, omit if the expression starts or ends with
# a bracket.
first = line.leaves[0]
second = line.leaves[1]
if first.type in OPENING_BRACKETS and second.type not in CLOSING_BRACKETS:
if _can_omit_opening_paren(line, first=first, line_length=line_length):
return True
# Note: we are not returning False here because a line might have *both*
# a leading opening bracket and a trailing closing bracket. If the
# opening bracket doesn't match our rule, maybe the closing will.
penultimate = line.leaves[-2]
last = line.leaves[-1]
if (
last.type == token.RPAR
or last.type == token.RBRACE
or (
# don't use indexing for omitting optional parentheses;
# it looks weird
last.type == token.RSQB
and last.parent
and last.parent.type != syms.trailer
)
):
if penultimate.type in OPENING_BRACKETS:
# Empty brackets don't help.
return False
if is_multiline_string(first):
# Additional wrapping of a multiline string in this situation is
# unnecessary.
return True
if _can_omit_closing_paren(line, last=last, line_length=line_length):
return True
return False
def _can_omit_opening_paren(line: Line, *, first: Leaf, line_length: int) -> bool:
"""See `can_omit_invisible_parens`."""
remainder = False
length = 4 * line.depth
_index = -1
for _index, leaf, leaf_length in line.enumerate_with_length():
if leaf.type in CLOSING_BRACKETS and leaf.opening_bracket is first:
remainder = True
if remainder:
length += leaf_length
if length > line_length:
break
if leaf.type in OPENING_BRACKETS:
# There are brackets we can further split on.
remainder = False
else:
# checked the entire string and line length wasn't exceeded
if len(line.leaves) == _index + 1:
return True
return False
def _can_omit_closing_paren(line: Line, *, last: Leaf, line_length: int) -> bool:
"""See `can_omit_invisible_parens`."""
length = 4 * line.depth
seen_other_brackets = False
for _index, leaf, leaf_length in line.enumerate_with_length():
length += leaf_length
if leaf is last.opening_bracket:
if seen_other_brackets or length <= line_length:
return True
elif leaf.type in OPENING_BRACKETS:
# There are brackets we can further split on.
seen_other_brackets = True
return False
def line_to_string(line: Line) -> str:
"""Returns the string representation of @line.
WARNING: This is known to be computationally expensive.
"""
return str(line).strip("\n")

View file

@ -0,0 +1,188 @@
"""Data structures configuring Black behavior.
Mostly around Python language feature support per version and Black configuration
chosen by the user.
"""
from hashlib import sha256
import sys
from dataclasses import dataclass, field
from enum import Enum, auto
from operator import attrgetter
from typing import Dict, Set
from warnings import warn
if sys.version_info < (3, 8):
from typing_extensions import Final
else:
from typing import Final
from black.const import DEFAULT_LINE_LENGTH
class TargetVersion(Enum):
PY33 = 3
PY34 = 4
PY35 = 5
PY36 = 6
PY37 = 7
PY38 = 8
PY39 = 9
PY310 = 10
class Feature(Enum):
F_STRINGS = 2
NUMERIC_UNDERSCORES = 3
TRAILING_COMMA_IN_CALL = 4
TRAILING_COMMA_IN_DEF = 5
# The following two feature-flags are mutually exclusive, and exactly one should be
# set for every version of python.
ASYNC_IDENTIFIERS = 6
ASYNC_KEYWORDS = 7
ASSIGNMENT_EXPRESSIONS = 8
POS_ONLY_ARGUMENTS = 9
RELAXED_DECORATORS = 10
PATTERN_MATCHING = 11
UNPACKING_ON_FLOW = 12
ANN_ASSIGN_EXTENDED_RHS = 13
FORCE_OPTIONAL_PARENTHESES = 50
# __future__ flags
FUTURE_ANNOTATIONS = 51
FUTURE_FLAG_TO_FEATURE: Final = {
"annotations": Feature.FUTURE_ANNOTATIONS,
}
VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = {
TargetVersion.PY33: {Feature.ASYNC_IDENTIFIERS},
TargetVersion.PY34: {Feature.ASYNC_IDENTIFIERS},
TargetVersion.PY35: {Feature.TRAILING_COMMA_IN_CALL, Feature.ASYNC_IDENTIFIERS},
TargetVersion.PY36: {
Feature.F_STRINGS,
Feature.NUMERIC_UNDERSCORES,
Feature.TRAILING_COMMA_IN_CALL,
Feature.TRAILING_COMMA_IN_DEF,
Feature.ASYNC_IDENTIFIERS,
},
TargetVersion.PY37: {
Feature.F_STRINGS,
Feature.NUMERIC_UNDERSCORES,
Feature.TRAILING_COMMA_IN_CALL,
Feature.TRAILING_COMMA_IN_DEF,
Feature.ASYNC_KEYWORDS,
Feature.FUTURE_ANNOTATIONS,
},
TargetVersion.PY38: {
Feature.F_STRINGS,
Feature.NUMERIC_UNDERSCORES,
Feature.TRAILING_COMMA_IN_CALL,
Feature.TRAILING_COMMA_IN_DEF,
Feature.ASYNC_KEYWORDS,
Feature.FUTURE_ANNOTATIONS,
Feature.ASSIGNMENT_EXPRESSIONS,
Feature.POS_ONLY_ARGUMENTS,
Feature.UNPACKING_ON_FLOW,
Feature.ANN_ASSIGN_EXTENDED_RHS,
},
TargetVersion.PY39: {
Feature.F_STRINGS,
Feature.NUMERIC_UNDERSCORES,
Feature.TRAILING_COMMA_IN_CALL,
Feature.TRAILING_COMMA_IN_DEF,
Feature.ASYNC_KEYWORDS,
Feature.FUTURE_ANNOTATIONS,
Feature.ASSIGNMENT_EXPRESSIONS,
Feature.RELAXED_DECORATORS,
Feature.POS_ONLY_ARGUMENTS,
Feature.UNPACKING_ON_FLOW,
Feature.ANN_ASSIGN_EXTENDED_RHS,
},
TargetVersion.PY310: {
Feature.F_STRINGS,
Feature.NUMERIC_UNDERSCORES,
Feature.TRAILING_COMMA_IN_CALL,
Feature.TRAILING_COMMA_IN_DEF,
Feature.ASYNC_KEYWORDS,
Feature.FUTURE_ANNOTATIONS,
Feature.ASSIGNMENT_EXPRESSIONS,
Feature.RELAXED_DECORATORS,
Feature.POS_ONLY_ARGUMENTS,
Feature.UNPACKING_ON_FLOW,
Feature.ANN_ASSIGN_EXTENDED_RHS,
Feature.PATTERN_MATCHING,
},
}
def supports_feature(target_versions: Set[TargetVersion], feature: Feature) -> bool:
return all(feature in VERSION_TO_FEATURES[version] for version in target_versions)
class Preview(Enum):
"""Individual preview style features."""
string_processing = auto()
remove_redundant_parens = auto()
one_element_subscript = auto()
class Deprecated(UserWarning):
"""Visible deprecation warning."""
@dataclass
class Mode:
target_versions: Set[TargetVersion] = field(default_factory=set)
line_length: int = DEFAULT_LINE_LENGTH
string_normalization: bool = True
is_pyi: bool = False
is_ipynb: bool = False
magic_trailing_comma: bool = True
experimental_string_processing: bool = False
python_cell_magics: Set[str] = field(default_factory=set)
preview: bool = False
def __post_init__(self) -> None:
if self.experimental_string_processing:
warn(
"`experimental string processing` has been included in `preview`"
" and deprecated. Use `preview` instead.",
Deprecated,
)
def __contains__(self, feature: Preview) -> bool:
"""
Provide `Preview.FEATURE in Mode` syntax that mirrors the ``preview`` flag.
The argument is not checked and features are not differentiated.
They only exist to make development easier by clarifying intent.
"""
if feature is Preview.string_processing:
return self.preview or self.experimental_string_processing
return self.preview
def get_cache_key(self) -> str:
if self.target_versions:
version_str = ",".join(
str(version.value)
for version in sorted(self.target_versions, key=attrgetter("value"))
)
else:
version_str = "-"
parts = [
version_str,
str(self.line_length),
str(int(self.string_normalization)),
str(int(self.is_pyi)),
str(int(self.is_ipynb)),
str(int(self.magic_trailing_comma)),
str(int(self.experimental_string_processing)),
str(int(self.preview)),
sha256((",".join(sorted(self.python_cell_magics))).encode()).hexdigest(),
]
return ".".join(parts)

View file

@ -0,0 +1,843 @@
"""
blib2to3 Node/Leaf transformation-related utility functions.
"""
import sys
from typing import (
Generic,
Iterator,
List,
Optional,
Set,
Tuple,
TypeVar,
Union,
)
if sys.version_info >= (3, 8):
from typing import Final
else:
from typing_extensions import Final
if sys.version_info >= (3, 10):
from typing import TypeGuard
else:
from typing_extensions import TypeGuard
from mypy_extensions import mypyc_attr
# lib2to3 fork
from blib2to3.pytree import Node, Leaf, type_repr, NL
from blib2to3 import pygram
from blib2to3.pgen2 import token
from black.cache import CACHE_DIR
from black.strings import has_triple_quotes
pygram.initialize(CACHE_DIR)
syms: Final = pygram.python_symbols
# types
T = TypeVar("T")
LN = Union[Leaf, Node]
LeafID = int
NodeType = int
WHITESPACE: Final = {token.DEDENT, token.INDENT, token.NEWLINE}
STATEMENT: Final = {
syms.if_stmt,
syms.while_stmt,
syms.for_stmt,
syms.try_stmt,
syms.except_clause,
syms.with_stmt,
syms.funcdef,
syms.classdef,
syms.match_stmt,
syms.case_block,
}
STANDALONE_COMMENT: Final = 153
token.tok_name[STANDALONE_COMMENT] = "STANDALONE_COMMENT"
LOGIC_OPERATORS: Final = {"and", "or"}
COMPARATORS: Final = {
token.LESS,
token.GREATER,
token.EQEQUAL,
token.NOTEQUAL,
token.LESSEQUAL,
token.GREATEREQUAL,
}
MATH_OPERATORS: Final = {
token.VBAR,
token.CIRCUMFLEX,
token.AMPER,
token.LEFTSHIFT,
token.RIGHTSHIFT,
token.PLUS,
token.MINUS,
token.STAR,
token.SLASH,
token.DOUBLESLASH,
token.PERCENT,
token.AT,
token.TILDE,
token.DOUBLESTAR,
}
STARS: Final = {token.STAR, token.DOUBLESTAR}
VARARGS_SPECIALS: Final = STARS | {token.SLASH}
VARARGS_PARENTS: Final = {
syms.arglist,
syms.argument, # double star in arglist
syms.trailer, # single argument to call
syms.typedargslist,
syms.varargslist, # lambdas
}
UNPACKING_PARENTS: Final = {
syms.atom, # single element of a list or set literal
syms.dictsetmaker,
syms.listmaker,
syms.testlist_gexp,
syms.testlist_star_expr,
syms.subject_expr,
syms.pattern,
}
TEST_DESCENDANTS: Final = {
syms.test,
syms.lambdef,
syms.or_test,
syms.and_test,
syms.not_test,
syms.comparison,
syms.star_expr,
syms.expr,
syms.xor_expr,
syms.and_expr,
syms.shift_expr,
syms.arith_expr,
syms.trailer,
syms.term,
syms.power,
}
ASSIGNMENTS: Final = {
"=",
"+=",
"-=",
"*=",
"@=",
"/=",
"%=",
"&=",
"|=",
"^=",
"<<=",
">>=",
"**=",
"//=",
}
IMPLICIT_TUPLE: Final = {syms.testlist, syms.testlist_star_expr, syms.exprlist}
BRACKET: Final = {
token.LPAR: token.RPAR,
token.LSQB: token.RSQB,
token.LBRACE: token.RBRACE,
}
OPENING_BRACKETS: Final = set(BRACKET.keys())
CLOSING_BRACKETS: Final = set(BRACKET.values())
BRACKETS: Final = OPENING_BRACKETS | CLOSING_BRACKETS
ALWAYS_NO_SPACE: Final = CLOSING_BRACKETS | {token.COMMA, STANDALONE_COMMENT}
RARROW = 55
@mypyc_attr(allow_interpreted_subclasses=True)
class Visitor(Generic[T]):
"""Basic lib2to3 visitor that yields things of type `T` on `visit()`."""
def visit(self, node: LN) -> Iterator[T]:
"""Main method to visit `node` and its children.
It tries to find a `visit_*()` method for the given `node.type`, like
`visit_simple_stmt` for Node objects or `visit_INDENT` for Leaf objects.
If no dedicated `visit_*()` method is found, chooses `visit_default()`
instead.
Then yields objects of type `T` from the selected visitor.
"""
if node.type < 256:
name = token.tok_name[node.type]
else:
name = str(type_repr(node.type))
# We explicitly branch on whether a visitor exists (instead of
# using self.visit_default as the default arg to getattr) in order
# to save needing to create a bound method object and so mypyc can
# generate a native call to visit_default.
visitf = getattr(self, f"visit_{name}", None)
if visitf:
yield from visitf(node)
else:
yield from self.visit_default(node)
def visit_default(self, node: LN) -> Iterator[T]:
"""Default `visit_*()` implementation. Recurses to children of `node`."""
if isinstance(node, Node):
for child in node.children:
yield from self.visit(child)
def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa: C901
"""Return whitespace prefix if needed for the given `leaf`.
`complex_subscript` signals whether the given leaf is part of a subscription
which has non-trivial arguments, like arithmetic expressions or function calls.
"""
NO: Final = ""
SPACE: Final = " "
DOUBLESPACE: Final = " "
t = leaf.type
p = leaf.parent
v = leaf.value
if t in ALWAYS_NO_SPACE:
return NO
if t == token.COMMENT:
return DOUBLESPACE
assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}"
if t == token.COLON and p.type not in {
syms.subscript,
syms.subscriptlist,
syms.sliceop,
}:
return NO
prev = leaf.prev_sibling
if not prev:
prevp = preceding_leaf(p)
if not prevp or prevp.type in OPENING_BRACKETS:
return NO
if t == token.COLON:
if prevp.type == token.COLON:
return NO
elif prevp.type != token.COMMA and not complex_subscript:
return NO
return SPACE
if prevp.type == token.EQUAL:
if prevp.parent:
if prevp.parent.type in {
syms.arglist,
syms.argument,
syms.parameters,
syms.varargslist,
}:
return NO
elif prevp.parent.type == syms.typedargslist:
# A bit hacky: if the equal sign has whitespace, it means we
# previously found it's a typed argument. So, we're using
# that, too.
return prevp.prefix
elif prevp.type in VARARGS_SPECIALS:
if is_vararg(prevp, within=VARARGS_PARENTS | UNPACKING_PARENTS):
return NO
elif prevp.type == token.COLON:
if prevp.parent and prevp.parent.type in {syms.subscript, syms.sliceop}:
return SPACE if complex_subscript else NO
elif (
prevp.parent
and prevp.parent.type == syms.factor
and prevp.type in MATH_OPERATORS
):
return NO
elif prevp.type == token.AT and p.parent and p.parent.type == syms.decorator:
# no space in decorators
return NO
elif prev.type in OPENING_BRACKETS:
return NO
if p.type in {syms.parameters, syms.arglist}:
# untyped function signatures or calls
if not prev or prev.type != token.COMMA:
return NO
elif p.type == syms.varargslist:
# lambdas
if prev and prev.type != token.COMMA:
return NO
elif p.type == syms.typedargslist:
# typed function signatures
if not prev:
return NO
if t == token.EQUAL:
if prev.type != syms.tname:
return NO
elif prev.type == token.EQUAL:
# A bit hacky: if the equal sign has whitespace, it means we
# previously found it's a typed argument. So, we're using that, too.
return prev.prefix
elif prev.type != token.COMMA:
return NO
elif p.type == syms.tname:
# type names
if not prev:
prevp = preceding_leaf(p)
if not prevp or prevp.type != token.COMMA:
return NO
elif p.type == syms.trailer:
# attributes and calls
if t == token.LPAR or t == token.RPAR:
return NO
if not prev:
if t == token.DOT or t == token.LSQB:
return NO
elif prev.type != token.COMMA:
return NO
elif p.type == syms.argument:
# single argument
if t == token.EQUAL:
return NO
if not prev:
prevp = preceding_leaf(p)
if not prevp or prevp.type == token.LPAR:
return NO
elif prev.type in {token.EQUAL} | VARARGS_SPECIALS:
return NO
elif p.type == syms.decorator:
# decorators
return NO
elif p.type == syms.dotted_name:
if prev:
return NO
prevp = preceding_leaf(p)
if not prevp or prevp.type == token.AT or prevp.type == token.DOT:
return NO
elif p.type == syms.classdef:
if t == token.LPAR:
return NO
if prev and prev.type == token.LPAR:
return NO
elif p.type in {syms.subscript, syms.sliceop}:
# indexing
if not prev:
assert p.parent is not None, "subscripts are always parented"
if p.parent.type == syms.subscriptlist:
return SPACE
return NO
elif not complex_subscript:
return NO
elif p.type == syms.atom:
if prev and t == token.DOT:
# dots, but not the first one.
return NO
elif p.type == syms.dictsetmaker:
# dict unpacking
if prev and prev.type == token.DOUBLESTAR:
return NO
elif p.type in {syms.factor, syms.star_expr}:
# unary ops
if not prev:
prevp = preceding_leaf(p)
if not prevp or prevp.type in OPENING_BRACKETS:
return NO
prevp_parent = prevp.parent
assert prevp_parent is not None
if prevp.type == token.COLON and prevp_parent.type in {
syms.subscript,
syms.sliceop,
}:
return NO
elif prevp.type == token.EQUAL and prevp_parent.type == syms.argument:
return NO
elif t in {token.NAME, token.NUMBER, token.STRING}:
return NO
elif p.type == syms.import_from:
if t == token.DOT:
if prev and prev.type == token.DOT:
return NO
elif t == token.NAME:
if v == "import":
return SPACE
if prev and prev.type == token.DOT:
return NO
elif p.type == syms.sliceop:
return NO
return SPACE
def preceding_leaf(node: Optional[LN]) -> Optional[Leaf]:
"""Return the first leaf that precedes `node`, if any."""
while node:
res = node.prev_sibling
if res:
if isinstance(res, Leaf):
return res
try:
return list(res.leaves())[-1]
except IndexError:
return None
node = node.parent
return None
def prev_siblings_are(node: Optional[LN], tokens: List[Optional[NodeType]]) -> bool:
"""Return if the `node` and its previous siblings match types against the provided
list of tokens; the provided `node`has its type matched against the last element in
the list. `None` can be used as the first element to declare that the start of the
list is anchored at the start of its parent's children."""
if not tokens:
return True
if tokens[-1] is None:
return node is None
if not node:
return False
if node.type != tokens[-1]:
return False
return prev_siblings_are(node.prev_sibling, tokens[:-1])
def parent_type(node: Optional[LN]) -> Optional[NodeType]:
"""
Returns:
@node.parent.type, if @node is not None and has a parent.
OR
None, otherwise.
"""
if node is None or node.parent is None:
return None
return node.parent.type
def child_towards(ancestor: Node, descendant: LN) -> Optional[LN]:
"""Return the child of `ancestor` that contains `descendant`."""
node: Optional[LN] = descendant
while node and node.parent != ancestor:
node = node.parent
return node
def replace_child(old_child: LN, new_child: LN) -> None:
"""
Side Effects:
* If @old_child.parent is set, replace @old_child with @new_child in
@old_child's underlying Node structure.
OR
* Otherwise, this function does nothing.
"""
parent = old_child.parent
if not parent:
return
child_idx = old_child.remove()
if child_idx is not None:
parent.insert_child(child_idx, new_child)
def container_of(leaf: Leaf) -> LN:
"""Return `leaf` or one of its ancestors that is the topmost container of it.
By "container" we mean a node where `leaf` is the very first child.
"""
same_prefix = leaf.prefix
container: LN = leaf
while container:
parent = container.parent
if parent is None:
break
if parent.children[0].prefix != same_prefix:
break
if parent.type == syms.file_input:
break
if parent.prev_sibling is not None and parent.prev_sibling.type in BRACKETS:
break
container = parent
return container
def first_leaf_column(node: Node) -> Optional[int]:
"""Returns the column of the first leaf child of a node."""
for child in node.children:
if isinstance(child, Leaf):
return child.column
return None
def is_arith_like(node: LN) -> bool:
"""Whether node is an arithmetic or a binary arithmetic expression"""
return node.type in {
syms.arith_expr,
syms.shift_expr,
syms.xor_expr,
syms.and_expr,
}
def is_docstring(leaf: Leaf) -> bool:
if prev_siblings_are(
leaf.parent, [None, token.NEWLINE, token.INDENT, syms.simple_stmt]
):
return True
# Multiline docstring on the same line as the `def`.
if prev_siblings_are(leaf.parent, [syms.parameters, token.COLON, syms.simple_stmt]):
# `syms.parameters` is only used in funcdefs and async_funcdefs in the Python
# grammar. We're safe to return True without further checks.
return True
return False
def is_empty_tuple(node: LN) -> bool:
"""Return True if `node` holds an empty tuple."""
return (
node.type == syms.atom
and len(node.children) == 2
and node.children[0].type == token.LPAR
and node.children[1].type == token.RPAR
)
def is_one_tuple(node: LN) -> bool:
"""Return True if `node` holds a tuple with one element, with or without parens."""
if node.type == syms.atom:
gexp = unwrap_singleton_parenthesis(node)
if gexp is None or gexp.type != syms.testlist_gexp:
return False
return len(gexp.children) == 2 and gexp.children[1].type == token.COMMA
return (
node.type in IMPLICIT_TUPLE
and len(node.children) == 2
and node.children[1].type == token.COMMA
)
def is_one_sequence_between(
opening: Leaf,
closing: Leaf,
leaves: List[Leaf],
brackets: Tuple[int, int] = (token.LPAR, token.RPAR),
) -> bool:
"""Return True if content between `opening` and `closing` is a one-sequence."""
if (opening.type, closing.type) != brackets:
return False
depth = closing.bracket_depth + 1
for _opening_index, leaf in enumerate(leaves):
if leaf is opening:
break
else:
raise LookupError("Opening paren not found in `leaves`")
commas = 0
_opening_index += 1
for leaf in leaves[_opening_index:]:
if leaf is closing:
break
bracket_depth = leaf.bracket_depth
if bracket_depth == depth and leaf.type == token.COMMA:
commas += 1
if leaf.parent and leaf.parent.type in {
syms.arglist,
syms.typedargslist,
}:
commas += 1
break
return commas < 2
def is_walrus_assignment(node: LN) -> bool:
"""Return True iff `node` is of the shape ( test := test )"""
inner = unwrap_singleton_parenthesis(node)
return inner is not None and inner.type == syms.namedexpr_test
def is_simple_decorator_trailer(node: LN, last: bool = False) -> bool:
"""Return True iff `node` is a trailer valid in a simple decorator"""
return node.type == syms.trailer and (
(
len(node.children) == 2
and node.children[0].type == token.DOT
and node.children[1].type == token.NAME
)
# last trailer can be an argument-less parentheses pair
or (
last
and len(node.children) == 2
and node.children[0].type == token.LPAR
and node.children[1].type == token.RPAR
)
# last trailer can be arguments
or (
last
and len(node.children) == 3
and node.children[0].type == token.LPAR
# and node.children[1].type == syms.argument
and node.children[2].type == token.RPAR
)
)
def is_simple_decorator_expression(node: LN) -> bool:
"""Return True iff `node` could be a 'dotted name' decorator
This function takes the node of the 'namedexpr_test' of the new decorator
grammar and test if it would be valid under the old decorator grammar.
The old grammar was: decorator: @ dotted_name [arguments] NEWLINE
The new grammar is : decorator: @ namedexpr_test NEWLINE
"""
if node.type == token.NAME:
return True
if node.type == syms.power:
if node.children:
return (
node.children[0].type == token.NAME
and all(map(is_simple_decorator_trailer, node.children[1:-1]))
and (
len(node.children) < 2
or is_simple_decorator_trailer(node.children[-1], last=True)
)
)
return False
def is_yield(node: LN) -> bool:
"""Return True if `node` holds a `yield` or `yield from` expression."""
if node.type == syms.yield_expr:
return True
if is_name_token(node) and node.value == "yield":
return True
if node.type != syms.atom:
return False
if len(node.children) != 3:
return False
lpar, expr, rpar = node.children
if lpar.type == token.LPAR and rpar.type == token.RPAR:
return is_yield(expr)
return False
def is_vararg(leaf: Leaf, within: Set[NodeType]) -> bool:
"""Return True if `leaf` is a star or double star in a vararg or kwarg.
If `within` includes VARARGS_PARENTS, this applies to function signatures.
If `within` includes UNPACKING_PARENTS, it applies to right hand-side
extended iterable unpacking (PEP 3132) and additional unpacking
generalizations (PEP 448).
"""
if leaf.type not in VARARGS_SPECIALS or not leaf.parent:
return False
p = leaf.parent
if p.type == syms.star_expr:
# Star expressions are also used as assignment targets in extended
# iterable unpacking (PEP 3132). See what its parent is instead.
if not p.parent:
return False
p = p.parent
return p.type in within
def is_multiline_string(leaf: Leaf) -> bool:
"""Return True if `leaf` is a multiline string that actually spans many lines."""
return has_triple_quotes(leaf.value) and "\n" in leaf.value
def is_stub_suite(node: Node) -> bool:
"""Return True if `node` is a suite with a stub body."""
if (
len(node.children) != 4
or node.children[0].type != token.NEWLINE
or node.children[1].type != token.INDENT
or node.children[3].type != token.DEDENT
):
return False
return is_stub_body(node.children[2])
def is_stub_body(node: LN) -> bool:
"""Return True if `node` is a simple statement containing an ellipsis."""
if not isinstance(node, Node) or node.type != syms.simple_stmt:
return False
if len(node.children) != 2:
return False
child = node.children[0]
return (
child.type == syms.atom
and len(child.children) == 3
and all(leaf == Leaf(token.DOT, ".") for leaf in child.children)
)
def is_atom_with_invisible_parens(node: LN) -> bool:
"""Given a `LN`, determines whether it's an atom `node` with invisible
parens. Useful in dedupe-ing and normalizing parens.
"""
if isinstance(node, Leaf) or node.type != syms.atom:
return False
first, last = node.children[0], node.children[-1]
return (
isinstance(first, Leaf)
and first.type == token.LPAR
and first.value == ""
and isinstance(last, Leaf)
and last.type == token.RPAR
and last.value == ""
)
def is_empty_par(leaf: Leaf) -> bool:
return is_empty_lpar(leaf) or is_empty_rpar(leaf)
def is_empty_lpar(leaf: Leaf) -> bool:
return leaf.type == token.LPAR and leaf.value == ""
def is_empty_rpar(leaf: Leaf) -> bool:
return leaf.type == token.RPAR and leaf.value == ""
def is_import(leaf: Leaf) -> bool:
"""Return True if the given leaf starts an import statement."""
p = leaf.parent
t = leaf.type
v = leaf.value
return bool(
t == token.NAME
and (
(v == "import" and p and p.type == syms.import_name)
or (v == "from" and p and p.type == syms.import_from)
)
)
def is_type_comment(leaf: Leaf, suffix: str = "") -> bool:
"""Return True if the given leaf is a special comment.
Only returns true for type comments for now."""
t = leaf.type
v = leaf.value
return t in {token.COMMENT, STANDALONE_COMMENT} and v.startswith("# type:" + suffix)
def wrap_in_parentheses(parent: Node, child: LN, *, visible: bool = True) -> None:
"""Wrap `child` in parentheses.
This replaces `child` with an atom holding the parentheses and the old
child. That requires moving the prefix.
If `visible` is False, the leaves will be valueless (and thus invisible).
"""
lpar = Leaf(token.LPAR, "(" if visible else "")
rpar = Leaf(token.RPAR, ")" if visible else "")
prefix = child.prefix
child.prefix = ""
index = child.remove() or 0
new_child = Node(syms.atom, [lpar, child, rpar])
new_child.prefix = prefix
parent.insert_child(index, new_child)
def unwrap_singleton_parenthesis(node: LN) -> Optional[LN]:
"""Returns `wrapped` if `node` is of the shape ( wrapped ).
Parenthesis can be optional. Returns None otherwise"""
if len(node.children) != 3:
return None
lpar, wrapped, rpar = node.children
if not (lpar.type == token.LPAR and rpar.type == token.RPAR):
return None
return wrapped
def ensure_visible(leaf: Leaf) -> None:
"""Make sure parentheses are visible.
They could be invisible as part of some statements (see
:func:`normalize_invisible_parens` and :func:`visit_import_from`).
"""
if leaf.type == token.LPAR:
leaf.value = "("
elif leaf.type == token.RPAR:
leaf.value = ")"
def is_name_token(nl: NL) -> TypeGuard[Leaf]:
return nl.type == token.NAME
def is_lpar_token(nl: NL) -> TypeGuard[Leaf]:
return nl.type == token.LPAR
def is_rpar_token(nl: NL) -> TypeGuard[Leaf]:
return nl.type == token.RPAR
def is_string_token(nl: NL) -> TypeGuard[Leaf]:
return nl.type == token.STRING

View file

@ -0,0 +1,60 @@
"""
Formatting numeric literals.
"""
from blib2to3.pytree import Leaf
def format_hex(text: str) -> str:
"""
Formats a hexadecimal string like "0x12B3"
"""
before, after = text[:2], text[2:]
return f"{before}{after.upper()}"
def format_scientific_notation(text: str) -> str:
"""Formats a numeric string utilizing scentific notation"""
before, after = text.split("e")
sign = ""
if after.startswith("-"):
after = after[1:]
sign = "-"
elif after.startswith("+"):
after = after[1:]
before = format_float_or_int_string(before)
return f"{before}e{sign}{after}"
def format_complex_number(text: str) -> str:
"""Formats a complex string like `10j`"""
number = text[:-1]
suffix = text[-1]
return f"{format_float_or_int_string(number)}{suffix}"
def format_float_or_int_string(text: str) -> str:
"""Formats a float string like "1.0"."""
if "." not in text:
return text
before, after = text.split(".")
return f"{before or 0}.{after or 0}"
def normalize_numeric_literal(leaf: Leaf) -> None:
"""Normalizes numeric (float, int, and complex) literals.
All letters used in the representation are normalized to lowercase."""
text = leaf.value.lower()
if text.startswith(("0o", "0b")):
# Leave octal and binary literals alone.
pass
elif text.startswith("0x"):
text = format_hex(text)
elif "e" in text:
text = format_scientific_notation(text)
elif text.endswith("j"):
text = format_complex_number(text)
else:
text = format_float_or_int_string(text)
leaf.value = text

View file

@ -0,0 +1,105 @@
"""Nice output for Black.
The double calls are for patching purposes in tests.
"""
import json
from typing import Any, Optional
from mypy_extensions import mypyc_attr
import tempfile
from click import echo, style
@mypyc_attr(patchable=True)
def _out(message: Optional[str] = None, nl: bool = True, **styles: Any) -> None:
if message is not None:
if "bold" not in styles:
styles["bold"] = True
message = style(message, **styles)
echo(message, nl=nl, err=True)
@mypyc_attr(patchable=True)
def _err(message: Optional[str] = None, nl: bool = True, **styles: Any) -> None:
if message is not None:
if "fg" not in styles:
styles["fg"] = "red"
message = style(message, **styles)
echo(message, nl=nl, err=True)
@mypyc_attr(patchable=True)
def out(message: Optional[str] = None, nl: bool = True, **styles: Any) -> None:
_out(message, nl=nl, **styles)
def err(message: Optional[str] = None, nl: bool = True, **styles: Any) -> None:
_err(message, nl=nl, **styles)
def ipynb_diff(a: str, b: str, a_name: str, b_name: str) -> str:
"""Return a unified diff string between each cell in notebooks `a` and `b`."""
a_nb = json.loads(a)
b_nb = json.loads(b)
diff_lines = [
diff(
"".join(a_nb["cells"][cell_number]["source"]) + "\n",
"".join(b_nb["cells"][cell_number]["source"]) + "\n",
f"{a_name}:cell_{cell_number}",
f"{b_name}:cell_{cell_number}",
)
for cell_number, cell in enumerate(a_nb["cells"])
if cell["cell_type"] == "code"
]
return "".join(diff_lines)
def diff(a: str, b: str, a_name: str, b_name: str) -> str:
"""Return a unified diff string between strings `a` and `b`."""
import difflib
a_lines = a.splitlines(keepends=True)
b_lines = b.splitlines(keepends=True)
diff_lines = []
for line in difflib.unified_diff(
a_lines, b_lines, fromfile=a_name, tofile=b_name, n=5
):
# Work around https://bugs.python.org/issue2142
# See:
# https://www.gnu.org/software/diffutils/manual/html_node/Incomplete-Lines.html
if line[-1] == "\n":
diff_lines.append(line)
else:
diff_lines.append(line + "\n")
diff_lines.append("\\ No newline at end of file\n")
return "".join(diff_lines)
def color_diff(contents: str) -> str:
"""Inject the ANSI color codes to the diff."""
lines = contents.split("\n")
for i, line in enumerate(lines):
if line.startswith("+++") or line.startswith("---"):
line = "\033[1m" + line + "\033[0m" # bold, reset
elif line.startswith("@@"):
line = "\033[36m" + line + "\033[0m" # cyan, reset
elif line.startswith("+"):
line = "\033[32m" + line + "\033[0m" # green, reset
elif line.startswith("-"):
line = "\033[31m" + line + "\033[0m" # red, reset
lines[i] = line
return "\n".join(lines)
@mypyc_attr(patchable=True)
def dump_to_file(*output: str, ensure_final_newline: bool = True) -> str:
"""Dump `output` to a temporary file. Return path to the file."""
with tempfile.NamedTemporaryFile(
mode="w", prefix="blk_", suffix=".log", delete=False, encoding="utf8"
) as f:
for lines in output:
f.write(lines)
if ensure_final_newline and lines and lines[-1] != "\n":
f.write("\n")
return f.name

View file

@ -0,0 +1,272 @@
"""
Parse Python code and perform AST validation.
"""
import ast
import platform
import sys
from typing import Any, Iterable, Iterator, List, Set, Tuple, Type, Union
if sys.version_info < (3, 8):
from typing_extensions import Final
else:
from typing import Final
# lib2to3 fork
from blib2to3.pytree import Node, Leaf
from blib2to3 import pygram
from blib2to3.pgen2 import driver
from blib2to3.pgen2.grammar import Grammar
from blib2to3.pgen2.parse import ParseError
from blib2to3.pgen2.tokenize import TokenError
from black.mode import TargetVersion, Feature, supports_feature
from black.nodes import syms
ast3: Any
_IS_PYPY = platform.python_implementation() == "PyPy"
try:
from typed_ast import ast3
except ImportError:
# Either our python version is too low, or we're on pypy
if sys.version_info < (3, 7) or (sys.version_info < (3, 8) and not _IS_PYPY):
print(
"The typed_ast package is required but not installed.\n"
"You can upgrade to Python 3.8+ or install typed_ast with\n"
"`python3 -m pip install typed-ast`.",
file=sys.stderr,
)
sys.exit(1)
else:
ast3 = ast
PY2_HINT: Final = "Python 2 support was removed in version 22.0."
class InvalidInput(ValueError):
"""Raised when input source code fails all parse attempts."""
def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]:
if not target_versions:
# No target_version specified, so try all grammars.
return [
# Python 3.7+
pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords,
# Python 3.0-3.6
pygram.python_grammar_no_print_statement_no_exec_statement,
# Python 3.10+
pygram.python_grammar_soft_keywords,
]
grammars = []
# If we have to parse both, try to parse async as a keyword first
if not supports_feature(
target_versions, Feature.ASYNC_IDENTIFIERS
) and not supports_feature(target_versions, Feature.PATTERN_MATCHING):
# Python 3.7-3.9
grammars.append(
pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords
)
if not supports_feature(target_versions, Feature.ASYNC_KEYWORDS):
# Python 3.0-3.6
grammars.append(pygram.python_grammar_no_print_statement_no_exec_statement)
if supports_feature(target_versions, Feature.PATTERN_MATCHING):
# Python 3.10+
grammars.append(pygram.python_grammar_soft_keywords)
# At least one of the above branches must have been taken, because every Python
# version has exactly one of the two 'ASYNC_*' flags
return grammars
def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -> Node:
"""Given a string with source, return the lib2to3 Node."""
if not src_txt.endswith("\n"):
src_txt += "\n"
grammars = get_grammars(set(target_versions))
errors = {}
for grammar in grammars:
drv = driver.Driver(grammar)
try:
result = drv.parse_string(src_txt, True)
break
except ParseError as pe:
lineno, column = pe.context[1]
lines = src_txt.splitlines()
try:
faulty_line = lines[lineno - 1]
except IndexError:
faulty_line = "<line number missing in source>"
errors[grammar.version] = InvalidInput(
f"Cannot parse: {lineno}:{column}: {faulty_line}"
)
except TokenError as te:
# In edge cases these are raised; and typically don't have a "faulty_line".
lineno, column = te.args[1]
errors[grammar.version] = InvalidInput(
f"Cannot parse: {lineno}:{column}: {te.args[0]}"
)
else:
# Choose the latest version when raising the actual parsing error.
assert len(errors) >= 1
exc = errors[max(errors)]
if matches_grammar(src_txt, pygram.python_grammar) or matches_grammar(
src_txt, pygram.python_grammar_no_print_statement
):
original_msg = exc.args[0]
msg = f"{original_msg}\n{PY2_HINT}"
raise InvalidInput(msg) from None
raise exc from None
if isinstance(result, Leaf):
result = Node(syms.file_input, [result])
return result
def matches_grammar(src_txt: str, grammar: Grammar) -> bool:
drv = driver.Driver(grammar)
try:
drv.parse_string(src_txt, True)
except (ParseError, TokenError, IndentationError):
return False
else:
return True
def lib2to3_unparse(node: Node) -> str:
"""Given a lib2to3 node, return its string representation."""
code = str(node)
return code
def parse_single_version(
src: str, version: Tuple[int, int]
) -> Union[ast.AST, ast3.AST]:
filename = "<unknown>"
# typed_ast is needed because of feature version limitations in the builtin ast
if sys.version_info >= (3, 8) and version >= (3,):
return ast.parse(src, filename, feature_version=version)
elif version >= (3,):
if _IS_PYPY:
return ast3.parse(src, filename)
else:
return ast3.parse(src, filename, feature_version=version[1])
raise AssertionError("INTERNAL ERROR: Tried parsing unsupported Python version!")
def parse_ast(src: str) -> Union[ast.AST, ast3.AST]:
# TODO: support Python 4+ ;)
versions = [(3, minor) for minor in range(3, sys.version_info[1] + 1)]
first_error = ""
for version in sorted(versions, reverse=True):
try:
return parse_single_version(src, version)
except SyntaxError as e:
if not first_error:
first_error = str(e)
raise SyntaxError(first_error)
ast3_AST: Final[Type[ast3.AST]] = ast3.AST
def _normalize(lineend: str, value: str) -> str:
# To normalize, we strip any leading and trailing space from
# each line...
stripped: List[str] = [i.strip() for i in value.splitlines()]
normalized = lineend.join(stripped)
# ...and remove any blank lines at the beginning and end of
# the whole string
return normalized.strip()
def stringify_ast(node: Union[ast.AST, ast3.AST], depth: int = 0) -> Iterator[str]:
"""Simple visitor generating strings to compare ASTs by content."""
node = fixup_ast_constants(node)
yield f"{' ' * depth}{node.__class__.__name__}("
type_ignore_classes: Tuple[Type[Any], ...]
for field in sorted(node._fields): # noqa: F402
# TypeIgnore will not be present using pypy < 3.8, so need for this
if not (_IS_PYPY and sys.version_info < (3, 8)):
# TypeIgnore has only one field 'lineno' which breaks this comparison
type_ignore_classes = (ast3.TypeIgnore,)
if sys.version_info >= (3, 8):
type_ignore_classes += (ast.TypeIgnore,)
if isinstance(node, type_ignore_classes):
break
try:
value: object = getattr(node, field)
except AttributeError:
continue
yield f"{' ' * (depth+1)}{field}="
if isinstance(value, list):
for item in value:
# Ignore nested tuples within del statements, because we may insert
# parentheses and they change the AST.
if (
field == "targets"
and isinstance(node, (ast.Delete, ast3.Delete))
and isinstance(item, (ast.Tuple, ast3.Tuple))
):
for elt in item.elts:
yield from stringify_ast(elt, depth + 2)
elif isinstance(item, (ast.AST, ast3.AST)):
yield from stringify_ast(item, depth + 2)
# Note that we are referencing the typed-ast ASTs via global variables and not
# direct module attribute accesses because that breaks mypyc. It's probably
# something to do with the ast3 variables being marked as Any leading
# mypy to think this branch is always taken, leaving the rest of the code
# unanalyzed. Tighting up the types for the typed-ast AST types avoids the
# mypyc crash.
elif isinstance(value, (ast.AST, ast3_AST)):
yield from stringify_ast(value, depth + 2)
else:
normalized: object
# Constant strings may be indented across newlines, if they are
# docstrings; fold spaces after newlines when comparing. Similarly,
# trailing and leading space may be removed.
if (
isinstance(node, ast.Constant)
and field == "value"
and isinstance(value, str)
):
normalized = _normalize("\n", value)
else:
normalized = value
yield f"{' ' * (depth+2)}{normalized!r}, # {value.__class__.__name__}"
yield f"{' ' * depth}) # /{node.__class__.__name__}"
def fixup_ast_constants(node: Union[ast.AST, ast3.AST]) -> Union[ast.AST, ast3.AST]:
"""Map ast nodes deprecated in 3.8 to Constant."""
if isinstance(node, (ast.Str, ast3.Str, ast.Bytes, ast3.Bytes)):
return ast.Constant(value=node.s)
if isinstance(node, (ast.Num, ast3.Num)):
return ast.Constant(value=node.n)
if isinstance(node, (ast.NameConstant, ast3.NameConstant)):
return ast.Constant(value=node.value)
return node

View file

@ -0,0 +1,106 @@
"""
Summarize Black runs to users.
"""
from dataclasses import dataclass
from enum import Enum
from pathlib import Path
from click import style
from black.output import out, err
class Changed(Enum):
NO = 0
CACHED = 1
YES = 2
class NothingChanged(UserWarning):
"""Raised when reformatted code is the same as source."""
@dataclass
class Report:
"""Provides a reformatting counter. Can be rendered with `str(report)`."""
check: bool = False
diff: bool = False
quiet: bool = False
verbose: bool = False
change_count: int = 0
same_count: int = 0
failure_count: int = 0
def done(self, src: Path, changed: Changed) -> None:
"""Increment the counter for successful reformatting. Write out a message."""
if changed is Changed.YES:
reformatted = "would reformat" if self.check or self.diff else "reformatted"
if self.verbose or not self.quiet:
out(f"{reformatted} {src}")
self.change_count += 1
else:
if self.verbose:
if changed is Changed.NO:
msg = f"{src} already well formatted, good job."
else:
msg = f"{src} wasn't modified on disk since last run."
out(msg, bold=False)
self.same_count += 1
def failed(self, src: Path, message: str) -> None:
"""Increment the counter for failed reformatting. Write out a message."""
err(f"error: cannot format {src}: {message}")
self.failure_count += 1
def path_ignored(self, path: Path, message: str) -> None:
if self.verbose:
out(f"{path} ignored: {message}", bold=False)
@property
def return_code(self) -> int:
"""Return the exit code that the app should use.
This considers the current state of changed files and failures:
- if there were any failures, return 123;
- if any files were changed and --check is being used, return 1;
- otherwise return 0.
"""
# According to http://tldp.org/LDP/abs/html/exitcodes.html starting with
# 126 we have special return codes reserved by the shell.
if self.failure_count:
return 123
elif self.change_count and self.check:
return 1
return 0
def __str__(self) -> str:
"""Render a color report of the current state.
Use `click.unstyle` to remove colors.
"""
if self.check or self.diff:
reformatted = "would be reformatted"
unchanged = "would be left unchanged"
failed = "would fail to reformat"
else:
reformatted = "reformatted"
unchanged = "left unchanged"
failed = "failed to reformat"
report = []
if self.change_count:
s = "s" if self.change_count > 1 else ""
report.append(
style(f"{self.change_count} file{s} ", bold=True, fg="blue")
+ style(f"{reformatted}", bold=True)
)
if self.same_count:
s = "s" if self.same_count > 1 else ""
report.append(style(f"{self.same_count} file{s} ", fg="blue") + unchanged)
if self.failure_count:
s = "s" if self.failure_count > 1 else ""
report.append(style(f"{self.failure_count} file{s} {failed}", fg="red"))
return ", ".join(report) + "."

View file

@ -0,0 +1,28 @@
"""An error-handling model influenced by that used by the Rust programming language
See https://doc.rust-lang.org/book/ch09-00-error-handling.html.
"""
from typing import Generic, TypeVar, Union
T = TypeVar("T")
E = TypeVar("E", bound=Exception)
class Ok(Generic[T]):
def __init__(self, value: T) -> None:
self._value = value
def ok(self) -> T:
return self._value
class Err(Generic[E]):
def __init__(self, e: E) -> None:
self._e = e
def err(self) -> E:
return self._e
Result = Union[Ok[T], Err[E]]

View file

@ -0,0 +1,238 @@
"""
Simple formatting on strings. Further string formatting code is in trans.py.
"""
import re
import sys
from functools import lru_cache
from typing import List, Pattern
if sys.version_info < (3, 8):
from typing_extensions import Final
else:
from typing import Final
STRING_PREFIX_CHARS: Final = "furbFURB" # All possible string prefix characters.
STRING_PREFIX_RE: Final = re.compile(
r"^([" + STRING_PREFIX_CHARS + r"]*)(.*)$", re.DOTALL
)
FIRST_NON_WHITESPACE_RE: Final = re.compile(r"\s*\t+\s*(\S)")
def sub_twice(regex: Pattern[str], replacement: str, original: str) -> str:
"""Replace `regex` with `replacement` twice on `original`.
This is used by string normalization to perform replaces on
overlapping matches.
"""
return regex.sub(replacement, regex.sub(replacement, original))
def has_triple_quotes(string: str) -> bool:
"""
Returns:
True iff @string starts with three quotation characters.
"""
raw_string = string.lstrip(STRING_PREFIX_CHARS)
return raw_string[:3] in {'"""', "'''"}
def lines_with_leading_tabs_expanded(s: str) -> List[str]:
"""
Splits string into lines and expands only leading tabs (following the normal
Python rules)
"""
lines = []
for line in s.splitlines():
# Find the index of the first non-whitespace character after a string of
# whitespace that includes at least one tab
match = FIRST_NON_WHITESPACE_RE.match(line)
if match:
first_non_whitespace_idx = match.start(1)
lines.append(
line[:first_non_whitespace_idx].expandtabs()
+ line[first_non_whitespace_idx:]
)
else:
lines.append(line)
return lines
def fix_docstring(docstring: str, prefix: str) -> str:
# https://www.python.org/dev/peps/pep-0257/#handling-docstring-indentation
if not docstring:
return ""
lines = lines_with_leading_tabs_expanded(docstring)
# Determine minimum indentation (first line doesn't count):
indent = sys.maxsize
for line in lines[1:]:
stripped = line.lstrip()
if stripped:
indent = min(indent, len(line) - len(stripped))
# Remove indentation (first line is special):
trimmed = [lines[0].strip()]
if indent < sys.maxsize:
last_line_idx = len(lines) - 2
for i, line in enumerate(lines[1:]):
stripped_line = line[indent:].rstrip()
if stripped_line or i == last_line_idx:
trimmed.append(prefix + stripped_line)
else:
trimmed.append("")
return "\n".join(trimmed)
def get_string_prefix(string: str) -> str:
"""
Pre-conditions:
* assert_is_leaf_string(@string)
Returns:
@string's prefix (e.g. '', 'r', 'f', or 'rf').
"""
assert_is_leaf_string(string)
prefix = ""
prefix_idx = 0
while string[prefix_idx] in STRING_PREFIX_CHARS:
prefix += string[prefix_idx]
prefix_idx += 1
return prefix
def assert_is_leaf_string(string: str) -> None:
"""
Checks the pre-condition that @string has the format that you would expect
of `leaf.value` where `leaf` is some Leaf such that `leaf.type ==
token.STRING`. A more precise description of the pre-conditions that are
checked are listed below.
Pre-conditions:
* @string starts with either ', ", <prefix>', or <prefix>" where
`set(<prefix>)` is some subset of `set(STRING_PREFIX_CHARS)`.
* @string ends with a quote character (' or ").
Raises:
AssertionError(...) if the pre-conditions listed above are not
satisfied.
"""
dquote_idx = string.find('"')
squote_idx = string.find("'")
if -1 in [dquote_idx, squote_idx]:
quote_idx = max(dquote_idx, squote_idx)
else:
quote_idx = min(squote_idx, dquote_idx)
assert (
0 <= quote_idx < len(string) - 1
), f"{string!r} is missing a starting quote character (' or \")."
assert string[-1] in (
"'",
'"',
), f"{string!r} is missing an ending quote character (' or \")."
assert set(string[:quote_idx]).issubset(
set(STRING_PREFIX_CHARS)
), f"{set(string[:quote_idx])} is NOT a subset of {set(STRING_PREFIX_CHARS)}."
def normalize_string_prefix(s: str) -> str:
"""Make all string prefixes lowercase."""
match = STRING_PREFIX_RE.match(s)
assert match is not None, f"failed to match string {s!r}"
orig_prefix = match.group(1)
new_prefix = (
orig_prefix.replace("F", "f")
.replace("B", "b")
.replace("U", "")
.replace("u", "")
)
# Python syntax guarantees max 2 prefixes and that one of them is "r"
if len(new_prefix) == 2 and "r" != new_prefix[0].lower():
new_prefix = new_prefix[::-1]
return f"{new_prefix}{match.group(2)}"
# Re(gex) does actually cache patterns internally but this still improves
# performance on a long list literal of strings by 5-9% since lru_cache's
# caching overhead is much lower.
@lru_cache(maxsize=64)
def _cached_compile(pattern: str) -> Pattern[str]:
return re.compile(pattern)
def normalize_string_quotes(s: str) -> str:
"""Prefer double quotes but only if it doesn't cause more escaping.
Adds or removes backslashes as appropriate. Doesn't parse and fix
strings nested in f-strings.
"""
value = s.lstrip(STRING_PREFIX_CHARS)
if value[:3] == '"""':
return s
elif value[:3] == "'''":
orig_quote = "'''"
new_quote = '"""'
elif value[0] == '"':
orig_quote = '"'
new_quote = "'"
else:
orig_quote = "'"
new_quote = '"'
first_quote_pos = s.find(orig_quote)
if first_quote_pos == -1:
return s # There's an internal error
prefix = s[:first_quote_pos]
unescaped_new_quote = _cached_compile(rf"(([^\\]|^)(\\\\)*){new_quote}")
escaped_new_quote = _cached_compile(rf"([^\\]|^)\\((?:\\\\)*){new_quote}")
escaped_orig_quote = _cached_compile(rf"([^\\]|^)\\((?:\\\\)*){orig_quote}")
body = s[first_quote_pos + len(orig_quote) : -len(orig_quote)]
if "r" in prefix.casefold():
if unescaped_new_quote.search(body):
# There's at least one unescaped new_quote in this raw string
# so converting is impossible
return s
# Do not introduce or remove backslashes in raw strings
new_body = body
else:
# remove unnecessary escapes
new_body = sub_twice(escaped_new_quote, rf"\1\2{new_quote}", body)
if body != new_body:
# Consider the string without unnecessary escapes as the original
body = new_body
s = f"{prefix}{orig_quote}{body}{orig_quote}"
new_body = sub_twice(escaped_orig_quote, rf"\1\2{orig_quote}", new_body)
new_body = sub_twice(unescaped_new_quote, rf"\1\\{new_quote}", new_body)
if "f" in prefix.casefold():
matches = re.findall(
r"""
(?:(?<!\{)|^)\{ # start of the string or a non-{ followed by a single {
([^{].*?) # contents of the brackets except if begins with {{
\}(?:(?!\})|$) # A } followed by end of the string or a non-}
""",
new_body,
re.VERBOSE,
)
for m in matches:
if "\\" in str(m):
# Do not introduce backslashes in interpolated expressions
return s
if new_quote == '"""' and new_body[-1:] == '"':
# edge case:
new_body = new_body[:-1] + '\\"'
orig_escape_count = body.count("\\")
new_escape_count = new_body.count("\\")
if new_escape_count > orig_escape_count:
return s # Do not introduce more escaping
if new_escape_count == orig_escape_count and orig_quote == '"':
return s # Prefer double quotes
return f"{prefix}{new_quote}{new_body}{new_quote}"

File diff suppressed because it is too large Load diff